summaryrefslogtreecommitdiffstats
path: root/xlators/cluster
Commit message (Collapse)AuthorAgeFilesLines
* cluster/afr: Unwind with correct pre/post parent bufsPranith Kumar K2012-08-023-403/+147
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | RCA: In case of dir fops create, mknod, mkdir, link, symlink, rename if the fop fails on read-child then unwinds are happening with all-zero pre/post iatt-bufs. The bug occurs because the parent bufs are not saved if the response is not from read-child. Fix: Save the pre/post-bufs for the first response. If the response comes from read-child, overwrite whatever we have cached. Tests: Attached the mount process to gdb. Tested that the unwinds happen with proper pre/post iatt bufs in the following cases: 1) All success case 2) Failure on read-child 3) Failure on non-read-child 4) Failure on all children. Tested soft-link self-heal to test the change made in that. Tested errno ENOTEMPTY for rmdir, rename fops. Change-Id: I82882423d2d766b4f4a3044203bcb5dbcaee1755 BUG: 845242 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3775 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Handle child_up & fd not opened case in xactionPranith Kumar K2012-08-011-7/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | RCA: When an fd is opened while a brick is down, after the brick comes back up afr issues open on the other brick. It can fail for a number of reasons (enoent etc). While the system is in that state, inode/entrylks pre-op happen only on the brick that is up and fd is opened for fd-fops. post-op should consider only the bricks where both pre-op and fop succeeded as success, rest of them as failures. Code now marks only the children that are down as failures as opposed to child_down & fd-not-opened. This makes change-log appear as success on the subvolume where we did not do any fop leading to no change-log but differences in data/metadata for reg-files. Fix: Mark non-participants of fop as failure. This is tracked in transaction.pre_op[]. Tests: Simulated the scenario using err-gen on top of one of the client xlator which fails all fops always. Performed fops and the changelog represented pending fops on the brick with err-gen loaded. Tested the case of brick down and perform entry/metadata/data operations to confirm they still work as expected. Change-Id: I41905936126b19abba56ca581c0301a894507e1a BUG: 844987 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3765 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Handle failures in fop_cbk gracefullyPranith Kumar K2012-07-311-31/+46
| | | | | | | | | | | | | | | | | | | | | | | | | RCA: Afr crashes when a last fop response fails and 'fop output' arguments are NULL. Afr does not handle these gracefully. Fix: Changed the fops to not access the 'fop output' arguments in case of failures. Tests: Changed afr wind_cbk code to fail the last response by setting op_ret as -1 and op_errno as ENOMEM and setting all other output variables as NULL to test the change. Removed the code to verify success cases. No crashes or errors seen. Change-Id: Iad9bc54db093a162f85bfb8dbeeda5b95acd21d8 BUG: 844689 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3760 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: update loc inode after inode_linkPranith Kumar K2012-07-311-12/+24
| | | | | | | | | | | | | | | | | | | | | | | | RCA: inode passed to inode_link is not assigned any gfid if the inode with that gfid is already linked, so loc for opendir does not have a valid inode Fix: Use the linked_inode returned by inode_link in the loc to perform further operations on the entry. Tests: Checked that opendir comes with an loc with valid inode. Checked that re-opendir happens successfully. Tested index, full self-heal work fine with the fix. Change-Id: Idf4ced4cc2320133744962059d363e373af0e5ec BUG: 826580 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3748 Reviewed-by: Raghavendra Bhat <raghavendra@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/stripe: handle short writes and errors in writev callbackBrian Foster2012-07-303-38/+106
| | | | | | | | | | | | | | | | | | | | | | | cluster/stripe write callback handling is broken in the event of server side errors and short writes due to crudely summing up the return values from each node. This can produce incorrect results or cause an application to rewrite the wrong portions of a buffer in an attempt to handle this condition. Modify cluster/stripe writev handling to record the requested size of each write and use this data to return the number of consecutive bytes written from the original request. This allows an application to retry a write at the point of error (and potentially consume said error). BUG: 809975 Change-Id: Ic35cb1e092c29545205aa32e352485c507534ce0 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3700 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Shishir Gowda <sgowda@redhat.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Modified split-brain handlingPranith Kumar K2012-07-265-57/+146
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | RCA The bug is observed because the decision to mark a file in split-brain is taken outside appropriate locks. Lookup gathers xattrs outside any lock. The xattrs being in split-brain in lookup should only be taken as a hint. Appropriate inodelks should be taken before confirming a split-brain. Self-heal confirms this at the moment. If data/metadata self-heal is turned off, inspecting of xattrs could not be performed so split-brain behavior does not work correctly if the self-heal options are turned off. Fix Self-heals are launched to inspect xattrs even when the data/metadata self-heal options are turned off. The decision to heal data/metadata after the xattrs are inspected is based on whether the options are turned on/off. So decision to set/reset split-brain flag is taken inside appropriate locks. Testcases: tests 33-36 in https://github.com/pranithk/gluster-tests/blob/master/afr/self-heal.sh Change-Id: Ia8aeab08208b50c06609ad35a9d72f3d553ee343 BUG: 833727 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3626 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Filter O_TRUNC in afr-fix-openPranith Kumar K2012-07-251-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | RCA: When open was done while a brick is down, afr opens the file after the brick comes backup. If this happens after the self-heal on the file is completed by self-heald etc, the file will end up in truncated state. Fix: Filter O_TRUNC while afr-fix-open because afr_open turns O_TRUNC into truncate transaction, so there will be pending changelog for the subvolume on which open fails. Testing: Had to simulate the race by stopping fix-open until self-heald completes self-heal on the file after brick online. Change-Id: I32759cc37f4bb34f206d01606a279f17b246dba4 BUG: 841840 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3705 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster: fix crash on link of named pipe in stripe/replicate volBrian Foster2012-07-252-12/+11
| | | | | | | | | | | | | | | | | | | | | | | | A crash occurs when attempting to link a named pipe on a striped, replicated volume. The cause for this crash is attempting to deref a NULL inode pointer in stripe_link_cbk(). The RCA for this bug uncovered a couple of problems: - AFR ignores the inode pointer it receives on failure (returning NULL). - stripe assumes the inode pointer is valid on failure. Either one of these changes addresses the crash, but this patch includes both changes. AFR is modified to pass along the inode pointer it receives (which could still be NULL). stripe is modified to not assume the inode pointer is valid on fop failure. BUG: 842825 Change-Id: I9cb2cc918552620929c3ecbd69bc66d4635eafdc Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3727 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Perform data self-heal for non regular filesPranith Kumar K2012-07-255-217/+162
| | | | | | | | | | | | | | | | | | | | | | | | | RCA: Data self-heal for non regular files open the files and then proceeds using that fd. This approach does not work for symlinks because open on symlink opens the file resolved by it. Fix: If the file is not a regular file then perform self-heal using loc. It needs to get 'big' lock and then perform lookup to get changelog then erase data part of chagelog, then unlock. Test cases: Automated at https://github.com/pranithk/gluster-tests/blob/master/afr/special-file-self-heal-test.sh Change-Id: I924a922f5135872efe2cccf2e712ada082c5689f BUG: 811317 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3724 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/stripe: don't fail if no fctx on a non-regular fileBrian Foster2012-07-251-10/+14
| | | | | | | | | | | | | cluster/stripe broke directory rename. Only check for fctx on regular files. BUG: 842652 Change-Id: I8a1e7ff30d57c994082cb10471f610023713ee53 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3720 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/distribute: Suppress user xattr mismatch log messageshishir gowda2012-07-251-1/+1
| | | | | | | | | | | | | | | | | Changing the log-level to DEBUG. Xattr mismatch can occur when parallel setxattr's race, or when one of the bricks was down. A subsequent setxattr will fix the condition when all the subvols are up. In this case, the 'user.swift' xattr used by ufo was out of sync, but did not cause any other error. Change-Id: I6fdff78869b8ff72c305bbe122033e6c1d9d3cff BUG: 838197 Signed-off-by: shishir gowda <sgowda@redhat.com> Reviewed-on: http://review.gluster.com/3722 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Mohammed Junaid <junaid@redhat.com> Reviewed-by: Vijay Bellur <vbellur@redhat.com>
* afr: pass back xdata in createBrian Foster2012-07-231-2/+5
| | | | | | | | | | | | | | | A striped, replicated volume spits an error on file creation because stripe requires xdata to process stripe information and AFR isn't passing it back. This fix was suggested by Amar Tumballi. BUG: 842373 Change-Id: Ia7063590ca5e873d4a4e155989cf067e8a07501f Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3713 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* stripe: filter coalesce key in getxattr()/listxattr()Amar Tumballi2012-07-171-0/+3
| | | | | | | | | | | | as 'stripe-coalesce' is an internal key, no need to show it on top of the mount-point. Change-Id: Iab836e73d59c42774db8a2eee13fe3b0cd994bc9 Signed-off-by: Amar Tumballi <amarts@redhat.com> BUG: 801887 Reviewed-on: http://review.gluster.com/3680 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Shishir Gowda <sgowda@redhat.com>
* glusterfs_ctx_t: un-globalize the filesystem contextAnand Avati2012-07-171-6/+4
| | | | | | | | | | | | | | | | | | | | | | | | So far there has been a global glusterfs_ctx_t object which represents the running instance of the filesystem (client or server). It contains the various graphs, connection to the management daemon over which new graphs are obtained, calls stacks issued on this filesystem, and a bunch of such things. With the introduction of libgfapi, it is no more true that there will be only one filesystem context in a process. Applications can be written to use libgfapi and obtain serveral instances of different filesystems/volumes in the same process. This involves messy untangling of assumptions inside libglusterfs that there would only be one global glusterfs_ctx_t and offload that assumption to glusterfsd/ and cli/ (where it is true). Change-Id: Ifd7d1259428c26076140a5764a2dc7361694139c BUG: 839950 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.com/3678 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com>
* remove useless if-before-free (and free-like) functionsJim Meyering2012-07-1320-180/+88
| | | | | | | | | | | | See comments in http://bugzilla.redhat.com/839925 for the code to perform this change. Signed-off-by: Jim Meyering <meyering@redhat.com> BUG: 839925 Change-Id: I10e4ecff16c3749fe17c2831c516737e08a3205a Reviewed-on: http://review.gluster.com/3661 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* core: remove the unused files - round 2Amar Tumballi2012-07-126-6034/+0
| | | | | | | | | BUG: 764890 Change-Id: I3eb626eeaa2a09f0e248444f560c2a0eaf46c642 Signed-off-by: Amar Tumballi <amarts@redhat.com> Reviewed-on: http://review.gluster.com/3660 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* xlator options: remove overwritten data-self-heal initializerJim Meyering2012-07-111-1/+0
| | | | | | | | | | | | | In the struct volume_options, the "data-self-heal" .default_value = "" setting appeared before a setting of .default_value = "on". Remove the former. Change-Id: Ieddcc18f61581f9448d806cd8bf8eefaaf0118b9 BUG: 789278 Signed-off-by: Jim Meyering <meyering@redhat.com> Reviewed-on: http://review.gluster.com/3589 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: post-op-delay supportAnand Avati2012-07-046-2/+175
| | | | | | | | | | | | | | | | post-op-delay introduces an artificial delay between the OP and POST-OP-CHANGELOG phases of a write transaction to increase the probability of changelog-piggyback and eager-locking to work more efficiently. Also enable eager-locking by default. Change-Id: I865ca4b68512c44818719c7e388952f15d53e6c2 BUG: 836033 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.com/3621 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Pranith Kumar Karampuri <pranithk@gluster.com>
* cluster/afr: cleanup lk_owner and PID messAnand Avati2012-07-043-41/+28
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Historically PID (frame->root->pid) was used by the locks translator to identify a locker (and make decisions about which locks contend or cooperate/merge). Since the introduction of lock_owner parameter the usage of PID (for locks) was deprecated and is now unused. This patch nukes the usage of PID in AFR The usage of lk_owner has also ended up being a mess, because of the differentiation required between ->lk() and ->inodelk(), (->lk() needs to be identified by the process (roughly) and ->inodelk() needs to be identified by the transaction) and also because of optimizations like eager locking (locks are no more identified by the transaction as they now get inherited by the next transaction). The scheme (and technique) now is: - All FOPs (the third phase of the transaction) happen with the lk_owner which is set by the topmost layer (FUSE, NFS etc.) - All entrylks are issued with lk_owner set to the frame->root address. - Inodelks which will not be subject to eager locking are issued with lk_owner set to frame->root. - Inodelks which are subject to eager locking are issued with lk_owner set to the address of fd_t (which are the only type of frames which get subject to the eager locking optimization) - At the start of the transaction, the transaction frame's lk_owner is set to the either frame->root or fd_t (and never unmodified) depending on the type of transaction. - Just before the third phase (FOP phase) the set lk_owner is "saved" away and overwritten by the lk_owner submitted by the top layer (FUSE or NFS) - Right after the third phase, the saved lk_owner is "restored" to resume the transaction into the POST-OP and eventually UNLOCK using the same lk_owner which was used during the LOCK phase. Change-Id: I6ab8e4d6b65ae4185fa85ad3fded8e9188b2f929 BUG: 836033 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.com/3620 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Pranith Kumar Karampuri <pranithk@gluster.com>
* cluster/afr: add a new option to select read-subvol by indexJeff Darcy2012-07-031-0/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | read subvolume is a nice option to set prefred read child if you have a replication over 2 datacenter. if you have 2 datacenter and have a distributed replication where one set of servers are in datacenter one and the other (the replicated) are in the other datacenter read-subvolume it not very handy since it goes over name and the subvolume name is different for each replication pair. i added a new option called read-subvolume-index which take the number of the subvolume to choose. 0 fo first , 1 for second and so on subvolume in every replication. this option can now be used in the --xlator-option mount option to choose the prefered read child for all replication at once. For Example on all clients in datacenter one you can use --xlator-option=volumename-replication-*.read-subvolume-index=0 to prefer read from the servers in datacenter one. when you expand or shrink the volume no changes are needed to the client config since the wildcard will set this option automatic on reconfigure. Change-Id: I3b47432f77037c380ff4a6296636c6f8fc953db9 BUG: 837420 Original-author: domwo <glusterfs@wollina.de> Signed-off-by: domwo <glusterfs@wollina.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.com/3615 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: fix for read-subvolume option does not workJeff Darcy2012-07-031-2/+2
| | | | | | | | | | | | | | changed order of prevered read child in afr_select_read_child_from_policy when a read child is set over config option read-subvolume it shoudl be first to return Change-Id: I1c5a8171379bb2bad76f6653e9d68a9349d55142 BUG: 833750 Original-author: domwo <glusterfs@wollina.de> Signed-off-by: domwo <glusterfs@wollina.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.com/3614 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/dht: Remove dht dependency on glusterfsd-mgmtshishir gowda2012-06-291-1/+7
| | | | | | | | | | | | | | | glusterfs_ctx->notify can be used by any xlator to talk to glusterfsd-mgmt. Note- This is for any rpc communication initiated by the xlator, and not from glusterd. Change-Id: Ic0e4af106fe1e98d797ca621facda8839b87598a BUG: 835757 Signed-off-by: shishir gowda <sgowda@redhat.com> Reviewed-on: http://review.gluster.com/3618 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Don't reset split-brain when data-self-heal is offPranith Kumar K2012-06-193-2/+4
| | | | | | | | | BUG: 804606 Change-Id: I8cefcb6efa687fac4ad412403c085b3767218f72 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3586 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Treat EEXIST as success in hardlinkPranith Kumar K2012-06-141-0/+12
| | | | | | | | | BUG: 831151 Change-Id: I6ecc099cf5f3ae58b19dfb00ed0b3f9959e711e5 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3571 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vijay@gluster.com>
* cluster/stripe: implement the coalesce stripe file formatBrian Foster2012-06-073-78/+407
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The coalesce file format for cluster/stripe condenses the striped files to a contiguous layout. The elimination of holes in striped files eliminates space wasted via local filesystem preallocation heuristics and significantly improves read performance. Coalesce mode is implemented with a new 'coalesce' xlator option, which is user-configurable and disabled by default. The format of newly created files is marked with a new 'stripe-coalesce' xattr. Cluster/stripe handles/preserves the format of files regardless of the current mode of operation (i.e., a volume can simultaneously consist of coalesced and non-coalesced files). Files without the stripe-coalesce attribute are assumed to have the traditional format to provide backward compatibility. extras/stripe-merge: support traditional and coalesce stripe formats Update the stripe-merge recovery tool to handle the traditional and coalesced file formats. The format of the file is detected automatically (and verified) via the stripe-coalesce attributes. BUG: 801887 Change-Id: I682f0b4e819f496ddb68c9a01c4de4688280fdf8 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3282 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* libglusterfs : Fix validation for integer volume options.Kaushal M2012-06-071-0/+1
| | | | | | | | | | | | | | | | | | | | | Integer volume options which specified only the min value as 0, would not be validated during "volume set". The range check for an option happened only if both min and max were not 0. In the above case, even though a minium was specified, the range check did not happen as both min and max were 0. To allow forced validation in such cases, a new member, "validate", has been added to volume_options_t. This member takes the values GF_OPT_VALIDATE_BOTH, GF_OPT_VALIDATE_MIN and GF_OPT_VALIDATE_MAX (GF_OPT_VALIDATE_BOTH is the default). Change-Id: I351de0eedb6028120e5c0b073ee5d9c141dee717 BUG: 809847 Signed-off-by: Kaushal M <kaushal@redhat.com> Reviewed-on: http://review.gluster.com/3084 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Check for null gfid_reqPranith Kumar K2012-06-061-1/+1
| | | | | | | | | | | | | | gfid_req is set only by the fuse xlator. Fresh lookups performed by self-heal-daemon, rebalance will not have gfid at all. Change-Id: I6712e3063067ecc5f19956e75d28c86bfc19fc65 BUG: 829203 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3529 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* core: coverity fixes (mostly resource leak fixes)Amar Tumballi2012-06-054-16/+42
| | | | | | | | | | | currently working on obvious resource leak reports in coverity Change-Id: I261f4c578987b16da399ab5a504ad0fda0b176b1 Signed-off-by: Amar Tumballi <amarts@redhat.com> BUG: 789278 Reviewed-on: http://review.gluster.com/3265 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* replicate: default read_child to a local brick if there is one.Jeff Darcy2012-06-055-3/+162
| | | | | | | | | | | Controlled by the "choose-local" option (on by default). Change-Id: I560f27c81703f2c9c62fdb51532c8eb763826df7 BUG: 806462 Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.com/3005 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Self-heald: inode_link files while crawlingPranith Kumar K2012-06-051-5/+10
| | | | | | | | | Change-Id: I559a3ff507b9487b1dfca7871c188a05d89ea6d6 BUG: 826580 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3515 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/dht: set conf->defrag to NULL after freeing the defrag structureRaghavendra Bhat2012-05-311-2/+3
| | | | | | | | | | | | | Also no need to free the xlator object after rebalance is over, as the process is about to be killed. Change-Id: I6973e43c0353b5de61c0b39e52a22c618be361f4 BUG: 826584 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3495 Reviewed-by: Amar Tumballi <amarts@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/distribute: set the inode layout in readdirp_cbk() for filesAmar Tumballi2012-05-311-0/+16
| | | | | | | | | | | | with this, inode-linking it in readdirp_cbk will be neater. Change-Id: Ie2cd646438f851e1755e9b6a3fc9898059bee359 Signed-off-by: Amar Tumballi <amar@gluster.com> BUG: 816140 Reviewed-on: http://review.gluster.com/2717 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* replicate: add hashed read-child method.Jeff Darcy2012-05-316-18/+93
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Both the first-to-respond method and the round-robin method are susceptible to clients repeatedly choosing the same servers across a series of opens, creating hot spots. Also, the code to handle a replica being down will ignore both methods and just choose the first remaining (which is not an issue for two-way but can be otherwise). The hashed method more reliably avoids such hot spots. There are three values/modes. 0: use the old (broken) methods. 1: select a read-child based on a hash of the file's GFID, so all clients will choose the same subvolume for a file (ensuring maximum consistency) but will distribute load for a set of files. 2: select a read-child based on a hash of the file's GFID plus the client's PID, so different children will distribute load even for one file. Mode 2 will probably be optimal for most cases. Using response time when we open the file is problematic, both because a single sample might not have been representative even then and because load might have shifted in the hours or days since (for long-lived files). Trying to use more current load information can lead to "herd following" behavior which is just as bad. Pseudo-random distribution is likely to be the best we can reasonably do, just as it is for DHT. Change-Id: I798c2760411eacf32e82a85f03bb7b08a4a49461 BUG: 802513 Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.com/2926 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* distribute: support user-specified layouts.Jeff Darcy2012-05-314-3/+15
| | | | | | | | | | | | | The new type is DHT_HASH_TYPE_DM_USER=1 (on disk in network byte order) and we treat it the same as DHT_HASH_TYPE_DM except that we don't stomp on it during rebalance. Change-Id: I893571a9b89577acdea2fe868915b18d3663fd77 BUG: 807312 Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.com/3004 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* storage/posix: Move landfill inside .glusterfsPranith Kumar K2012-05-313-41/+2
| | | | | | | | | Change-Id: Ia2944f891dd62e72f3c79678c3a1fed389854a90 BUG: 811970 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3158 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: decrement changelog when fdctx is absentPranith Kumar K2012-05-241-0/+2
| | | | | | | | | Change-Id: I9d76ddbd2cf8e4e8e4ad70529ba3a70178489a68 BUG: 765194 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3435 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* glusterd/rebalance: Display run-time of rebalance processshishir gowda2012-05-243-3/+59
| | | | | | | | | | | | | | | | | Added run-time value field to cli output of rebalance/remove-brick. A new cluster/distribute boolean option rebalance-stats when set to ON, time taken for migration of each file is logged. With rebalance-stats OFF (default), rebalance logs will only have entries showing time spent in each directory. Change-Id: I02a8918621120068cd71ffaf2999d30b3a2d10a2 BUG: 821987 Signed-off-by: shishir gowda <shishirng@gluster.com> Reviewed-on: http://review.gluster.com/3303 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Release inodelk on erase changelog failuresPranith Kumar K2012-05-231-0/+3
| | | | | | | | | Change-Id: I58271e1ac5a116b5bc717d7cad9f03eb7dc8a1a4 BUG: 811551 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3417 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* distribute: use global synctask 'env' instead of localAmar Tumballi2012-05-236-41/+3
| | | | | | | | | | | | | | creating a local synctask_env can lead to creating of many more syncop threads than required. The current syncop logic can handle the scale-up/scale-down of threads depending on the load. Hence, its neater to use global synctask env. Change-Id: Id46f963a0190c0154513317ae03323db155ac15a Signed-off-by: Amar Tumballi <amarts@redhat.com> BUG: 823774 Reviewed-on: http://review.gluster.com/3412 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Unlock higher entry locks in rename entrylk failure.Pranith Kumar K2012-05-221-0/+4
| | | | | | | | | BUG: 823255 Change-Id: Ic6ad33518ea42c9518a21381518bd4f4afdd87cb Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3382 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Thou shalt not free(3) memory dirname(3) returnedEmmanuel Dreyfus2012-05-211-1/+9
| | | | | | | | | | | | | | On Linux basename() and dirname() return a pointer within the string passed as argument. On BSD flavors, basename() and dirname() return static storage, or pthread specific storage. Both behaviour are compliant, but calling free on the result in the second case is a bug. BUG: 764655 Change-Id: Ic82414aff1f8db2a7544b16315761ce1c05276c4 Signed-off-by: Emmanuel Dreyfus <manu@netbsd.org> Reviewed-on: http://review.gluster.com/3377 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Pranith Kumar Karampuri <pranithk@gluster.com>
* cluster/afr: Assign gfid path if path is NULL in lookupPranith Kumar K2012-05-211-9/+14
| | | | | | | | | | | Change-Id: I45be4ea7f04ee79b67a83134fe8ebd18067a707f BUG: 820355 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3373 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Raghavendra Bhat <raghavendrabhat@gluster.com> Reviewed-by: Vijay Bellur <vijay@gluster.com>
* cluster/afr: Enforce order in pre/post opPranith Kumar K2012-05-189-100/+135
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The xattrop order in pre/post op on all the subvols is client-0, client-1... client-n where n is (replica-count - 1). This order can lead to invalid split-brains if the brick dies in the middle of xattrops. Example: transaction completed pre-op, so on all the subvolumes xattrs have '1' changelog. Now post-op is sent to both the subvols. On subvol-0 change-log of client-0 is decremented to 0, before decrementing change-log of client-1 to 0 the brick dies. This change-log status on subvol-0 gives the meaning that a change is done on subvol-0 successfully but on subvol-1 it failed. Which is not what happened. Changes done when the subvol-0 was down will lead to pending change-log on subvol-1 for subvol-0. Which is correct. When the subvol-0 is brought back up, the change-log will be in split-brain state even when it is not a legitimate split-brain. If the brick dies in the middle of xattrops it should remain fool. Pre-op should perform xattrop of the local change-log first and post-op should perform xattrop of the local change-log last. In case of optimistic changelogs txn_changelog should be done last on local if it succeeds, first if it fails. Change-Id: Ib6eeb20cdc49b0b1fd2f454f25a9c8e08388c6e7 BUG: 765194 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3226 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Perform conservative merge on dir with xattr split-brainPranith Kumar K2012-05-183-10/+61
| | | | | | | | | | Change-Id: I96d59ad239c2c5efee14dd4b01a10a3f565d491e BUG: 765587 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3091 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Handle files w.o. xattrs and size mismatch.Pranith Kumar K2012-05-182-10/+44
| | | | | | | | | | Change-Id: Ia27ee996bed8f5915c154718bf6e859b6a2fc335 BUG: 765587 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3090 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Mark zero size file as sink in absense of xattrs.Pranith Kumar K2012-05-182-2/+53
| | | | | | | | | | Change-Id: I4500f39a49ee16e6e88451dcf147d9f49b1d749e BUG: 765587 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3089 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* cluster/afr: Determining sources should do both fxattrop, fstatPranith Kumar K2012-05-182-95/+65
| | | | | | | | | Change-Id: Ifab37db2af8d489cd516e992b7423c765dcabc4f BUG: 765587 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3088 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Self-heald: Enable configuring of heal poll timeoutPranith Kumar K2012-05-183-3/+12
| | | | | | | | | Change-Id: I631e5bf4b3615b553b72e7ac7f490714b3b995f9 BUG: 821395 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3329 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Fixed a memory leak in stripe translator.Alexander Bersenev2012-05-171-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | When iobuf is created it has reference count = 1. After iobref_add (local->iobref, iobuf); reference count becomes 2. After iobref_unref(local->iobref); it becomes 1 and never becomes 0. So iobuf never deletes and this causes a memory leak. I emulated it, creating files on brick manually. After 5 mins of: while true; do dd if=file of=/dev/zero bs=16384; done top showed me this: 4618 root 20 0 1721m 1.5g 1868 S 0.0 16.2 5:41.77 glusterfs 1.5 gb of memory has leaked. For what this if for? Can it be true in the normal conditions? if ((local->replies[i].op_ret < local->replies[i].requested_size) && (local->stbuf_size > (local->offset + op_ret))) { May be delete it entirely? Change-Id: I17c115ab566e5bba662dd809e0c747db3c0310c8 BUG: 822378 Signed-off-by: Alexander Bersenev <bay@hackerdom.ru> Reviewed-on: http://review.gluster.com/3340 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Self-heald: Print the time stamp of the event-historyPranith Kumar K2012-05-171-10/+22
| | | | | | | | | Change-Id: I45767e26288ef6de6446ddf2ea82ed31e128d227 BUG: 796579 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3277 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vijay@gluster.com>
generated by cgit (git 2.34.1) at 2024-11-19 23:13:19 +0000