summaryrefslogtreecommitdiffstats
path: root/xlators/performance
Commit message (Collapse)AuthorAgeFilesLines
* cluster/afr: Avoid self-healing extended attribute used by SELinux.Vijay Bellur2013-04-301-1/+1
| | | | | | | | | | | | | | | | | | Since removexattr() fails to remove "security.selinux" in a system where SELinux is enforcing, xattr self-healing fails. As a consequence of this, user extended attributes are not being healed. Added a check in afr to prune SELinux xattr from the dictionary used for removing xattrs from the sink. Minor changes in tests and md-cache as well. Signed-off-by: Vijay Bellur <vbellur@redhat.com> Change-Id: I854bfc0098dde812ce2afe64b125ee40c04bdeb1 BUG: 957877 Reviewed-on: http://review.gluster.org/4905 Reviewed-by: Venky Shankar <vshankar@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/io-cache: Avoid double mem_put in ioc_readvPranith Kumar K2013-04-261-2/+3
| | | | | | | | | | | | | On readv error io-cache frame->local is not set to NULL so the local is mem_put in STACK_DESTROY as well. This patch sets frame->local to NULL in all cases. Change-Id: I00013df1377475aa5f3c0c681dcb58b32e1e8063 BUG: 955751 Signed-off-by: Pranith Kumar K <pkarampu@redhat.com> Reviewed-on: http://review.gluster.org/4884 Reviewed-by: Raghavendra G <raghavendra@gluster.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
* performance/io-threads: Fix range-check for least-rate-limitPranith Kumar K2013-03-211-0/+1
| | | | | | | | | | | | | The issue could be fixed with .validate=GF_OPT_VALIDATE_MIN. But adding max value is more robust. Change-Id: Ia69c6f86855dbd34a26e20391e77bfa0f796a200 BUG: 923573 Signed-off-by: Pranith Kumar K <pkarampu@redhat.com> Reviewed-on: http://review.gluster.org/4698 Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
* performance/write-behind: guarantee non-overlapping concurrent writesAnand Avati2013-02-201-1/+65
| | | | | | | | | | | | | | | | | Maintain a list of writes (either written behind or SYNC) which are currently "in progress" (i.e, STACK_WIND'ed towards server) and hold off any new STACK_WIND of write (either written behind or SYNC) which overlaps with any of the "in progress" writes. This is a guarantee which AFR's eager-lock depends upon (though not strictly a write-behind requirement) Change-Id: Icedd0b51b440366a906dc9223d62b7fd6ef2ca03 BUG: 857673 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4551 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Raghavendra G <raghavendra@gluster.com>
* read-ahead: re-enable support for variable page sizeAnand Avati2013-02-201-1/+12
| | | | | | | | | | | | | | | | Support for variable size page-size was disabled with the introduction of fixed size iobufs. Since the introduction of variable sized iobufs there is no reason to not have configurable page-size in read-ahead. This patch enables necessary changes in the translator for configurable page-size. Change-Id: I677d70fef50641eb041269aca92a088b9d4961cc BUG: 764204 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4526 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Raghavendra G <raghavendra@gluster.com>
* open-behind: propagate errors from ob_wake_cbkAnand Avati2013-02-201-9/+25
| | | | | | | | | | | | | | | If opening fd in background fails, then remember the error and fail all further calls on the fd. Use the newly introduced call_unwind_error() function from call-stub cleanup to fail the future calls. Change-Id: I3b09b7969c98d915abd56590a2777ce833b81813 BUG: 846240 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4521 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Raghavendra Bhat <raghavendra@redhat.com>
* call-stub: internal refactorAnand Avati2013-02-191-40/+40
| | | | | | | | | | | | | | | | | - re-structure members of call_stub_t with new simpler layout - easier to inspect call_stub_t contents in gdb now - fix a bunch of double unrefs and double frees in cbk stub - change all STACK_UNWIND to STACK_UNWIND_STRICT and thereby fixed a lot of bad params - implement new API call_unwind_error() which can even be called on fop_XXX_stub(), and not necessarily fop_XXX_cbk_stub() Change-Id: Idf979f14d46256af0afb9658915cc79de157b2d7 BUG: 846240 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4520 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
* performance/open-behind: use anonymous fd for doing fstat and readvRaghavendra Bhat2013-02-191-2/+2
| | | | | | | | | Change-Id: I61a3c221e0a15736ab6315e2538c03dac27480a5 BUG: 846240 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/4483 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: mark fd bad if any written behind writes fail.Raghavendra G2013-02-171-57/+114
| | | | | | | | | Change-Id: I515fc26c61e1ea929a3049b3001c58a64f3e6c87 BUG: 765473 Signed-off-by: Raghavendra G <raghavendra@gluster.com> Reviewed-on: http://review.gluster.org/4515 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* Use proper libtool option -avoid-version instead of bogus -avoidversionAnand Avati2013-02-078-8/+8
| | | | | | | | | | Change-Id: I1c9541058c7d07786539a3266ca125a6a15287d8 BUG: 859835 Signed-off-by: Anand Avati <avati@redhat.com> Original-author: Kacper Kowalik (Xarthisius) <xarthisius.kk@gmail.com> Signed-off-by: Kacper Kowalik (Xarthisius) <xarthisius.kk@gmail.com> Reviewed-on: http://review.gluster.org/3967 Tested-by: Gluster Build System <jenkins@build.gluster.com>
* open-behind: translator to perform open calls in backgroundAnand Avati2013-02-065-1/+958
| | | | | | | | | | | | | | | | | | | | This is functionality peeled out of quick-read into a separate translator. Fops which modify the file (where it is required to perform the operation on the true fd) will trigger and wait for the backend open to succeed and use that fd. Fops like fstat() readv() etc. will use anonymous FD (configurable) when original fd is unopened at the backend. Change-Id: Id9847fdbfdc82c1c8e956339156b6572539c1876 BUG: 846240 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4406 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
* performance/write-behind: do not try to take LOCK in forgetRaghavendra Bhat2013-02-031-7/+3
| | | | | | | | | | | LOCK attempt in wb_forget is unnecessary Change-Id: Ibdedc23d0c829c34aedd6fc5bc0e0a584b832514 BUG: 903566 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/4423 Reviewed-by: Anand Avati <avati@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
* quick-read: various fixesAnand Avati2013-01-291-1/+12
| | | | | | | | | | | | | | | | | | | | | | - initialize xdata in qr_lookup even if it was NULL from top. This allows qr to do its job even if lookup originated from fuse-resolve.c - extend test cases to include 1 second delay and retry - fix bug while checking condition for cached unwind qr_readv_cached() unwinds if op_ret > 0. Therefore qr_readv() must wind to subvol only if !(op_ret > 0) (i.e, op_ret <= 0). - qr_readv_cached() is using uninitialized @conf pointer. Thanks to Raghavendra Bhat for catching this! Change-Id: Ifaf2ea2685e452210ef9ba3c2d1f2ab51900650c BUG: 846240 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4452 Reviewed-by: Raghavendra Bhat <raghavendra@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
* performance/io-cache: propagate errors while unwinding frame inRaghavendra G2013-01-291-5/+11
| | | | | | | | | | | read path. Change-Id: Ieb5d592a987e8681d5ec019da309f75e3b207580 BUG: 858242 Signed-off-by: Raghavendra G <raghavendra@gluster.com> Reviewed-on: http://review.gluster.org/4204 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/md-cache: add force-readdirp flag to make readdirp configurableBrian Foster2013-01-281-3/+29
| | | | | | | | | | | | | | md-cache currently transforms all readdir fops into readdirp fops. This patch creates the 'force-readdirp' configuration flag to provide control over this behavior. force-readdirp is enabled by default to maintain current default behavior. BUG: 903175 Change-Id: Idd70926dec7c271204bdfb11fb052e56d0a39420 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.org/4440 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* quick-read: refactorAnand Avati2013-01-283-3340/+413
| | | | | | | | | | | | | | | | | - peel out 'open behind' functionality into a separate translator - issue where, if file size had grown by revalidate, data was not flushed - removed unnecessary acquistion of table->lock (e.g in qr_lookup()) - keep inode ctx persistent, prune only data (effectively changing the order of lock acquisition from INODE -> TABLE) - validation with readdirplus - use variable size iobufs to simply cached reads Change-Id: If1586d0298fd1697ddff9fd7008efb3d286d436a BUG: 846240 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4403 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
* "gcc -pedantic": made 'inline' functions as 'static inline' functionsAmar Tumballi2013-01-232-6/+4
| | | | | | | | | | | | for passing the build with -pedantic flag Change-Id: I80fd9528321e4c6ea5bec32bf5cdc54cc4e4f65e BUG: 875913 Signed-off-by: Amar Tumballi <amarts@redhat.com> Reviewed-on: http://review.gluster.org/4186 Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* glusterd/cli: Updated the options descriptions for "volume set help"Avra Sengupta2013-01-211-0/+2
| | | | | | | | | Change-Id: I0db00b7334bb9707ab48bd661ac03a3ad818d6e4 BUG: 893458 Signed-off-by: Avra Sengupta <asengupt@redhat.com> Reviewed-on: http://review.gluster.org/4393 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* core: fixes for gcc's '-pedantic' flag buildAvra Sengupta2013-01-212-3/+6
| | | | | | | | | | | | | * warnings on 'void *' arguments * warnings on empty initializations * warnings on empty array (array[0]) Change-Id: Iae440f54cbd59580eb69f3ecaed5a9926c0edf95 BUG: 875913 Signed-off-by: Avra Sengupta <asengupt@redhat.com> Reviewed-on: http://review.gluster.org/4219 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* write-behind: fixes issues with iobuf length for large writesKaushal M2012-12-131-5/+15
| | | | | | | | | | | | | | | | Use of an unsigned type in some calculations of size would lead to segmentation faults, if several large adjacent writes came in concurrently. Also, improves buffer allocation code to take the size required into account. Credits for the patch go to Amar. Change-Id: I8a09c52d49909e4ee8e7d4dcfa02ec33ea36a551 BUG: 880948 Signed-off-by: Kaushal M <kaushal@redhat.com> Reviewed-on: http://review.gluster.org/4307 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* fix memory leaksRaghavendra Bhat2012-12-041-0/+2
| | | | | | | | | | | | | * write-behind: free the inode context in wb_forget * distribute: in readdirp callback put the allocated context to the inode * distribute: check if the layout is NULL before accessing it in layout_unref Change-Id: I7698f81b85b99d06bf6b01fc1a6e51e1593b5e27 BUG: 790709 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/4250 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* perf/io-threads: least-rate-limit least priority throttlingBrian Foster2012-11-212-2/+94
| | | | | | | | | | | | | | | | | | | | | | | | | | | The 'least-rate-limit' io-threads translator option enables throttling of least priority operations. This is initially intended as a debug/diagnostic tool for users who might experience overloaded servers via background activity (i.e., self-heal). least-rate-limit defines the maximum number of least priority operations the io-threads translator will dequeue in one second. If the specified rate limit is met, the worker threads sleep for the minimal amount of time before the next least priority operation becomes available (or until a new request arrives). The requests/second metric is generic and relative to a variety of factors involved with a background operation (server, storage, etc.). The most recent measured rate ("cached least rate") is added to the io-threads state dump content (kill -USR1) to serve as a reference point to throttle background activity under particular conditions. Change-Id: I80f2282992137d57b1becaa5c6ae3858c066862a BUG: 853680 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.org/4119 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* write-behind: use uint64_t for overlap comparisonAnand Avati2012-10-131-4/+4
| | | | | | | | | off_t is 'long int' (signed word) and therefore ULLONG_MAX - 1 Change-Id: I027de7a1b2ca24865d5d787f9986930e97911ca4 BUG: 857673 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/4079
* performance/write-behind: use try lock while taking statedumpsRaghavendra Bhat2012-10-111-2/+9
| | | | | | | | | | Change-Id: I690e8bf650d6e6e50899c2e17a79f42789e701eb BUG: 843792 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/4036 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* build: split CPPFLAGS from CFLAGSJeff Darcy2012-10-037-14/+23
| | | | | | | | | | | | | | | | | Automake provides a separate variable for preprocessor flags (*_CPPFLAGS). They are already uses in a few places, so make it consistent and use it everywhere. Note that cflags obtained from pkg-config often are cppflags, which is why LIBXML2_CFLAGS moves with into AM_CPPFLAGS, for example. Change-Id: I15feed1d18b2ca497371271c4b5876d5ec6289dd BUG: 862082 Original-author: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.org/4029 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* build: remove useless explicit -fPIC -shared fromJeff Darcy2012-10-037-14/+14
| | | | | | | | | | | | | | | | | | | | CFLAGS libtool will automatically add "-fPIC" to the compiler command line as needed, so there is no need to specify it separately. "-shared" is normally a linker flag and has an odd effect when used with libtool --mode=compile, namely that it inhibits production of static objects. For that however, using AC_DISABLE_STATIC is a lot simpler. Change-Id: Ic4cba0fad18ffd985cf07f8d6951a976ae59a48f BUG: 862082 Original-author: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.org/4027 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* build: remove -nostartfiles flagJeff Darcy2012-10-027-7/+7
| | | | | | | | | | | | | | | The "-nostartfiles" is a discouraged option and is documented to potentially result in undesired behavior. Since I see no reason why it should be in glusterfs, remove it. Change-Id: I56f2b08874516ebad91447b2583ca2fb776bb7ab BUG: 862082 Original-author: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.org/4018 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* build: consolidate common compilation flags into one variableJeff Darcy2012-10-017-7/+7
| | | | | | | | | | | | | | | Some -D flags are present in all files, so collect them. This adds -D${GF_HOST_OS} to some compiler command lines, but this should not be a problem. Change-Id: I1aeb346143d4984c9cc4f2750c465ce09af1e6ca BUG: 862082 Original-author: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jan Engelhardt <jengelh@inai.de> Signed-off-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-on: http://review.gluster.org/4013 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* write-behind: implement causal ordering and other cleanupAnand Avati2012-10-011-2418/+1068
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Rules of causal ordering implemented: - If request A arrives after the acknowledgement (to the app, i.e, STACK_UNWIND) of another request B, then request B is said to have 'caused' request A. - (corollary) Two requests, which at any point of time, are unacknowledged simultaneously in the system can never 'cause' each other (wb_inode->gen is based on this) - If request A is caused by request B, AND request A's region has an overlap with request B's region, then then the fulfillment of request A is guaranteed to happen after the fulfillment of B. - FD of origin is not considered for the determination of causal ordering. - Append operation's region is considered the whole file. Other cleanup: - wb_file_t not required any more. - wb_local_t not required any more. - O_RDONLY fd's operations now go through the queue to make sure writes in the requested region get fulfilled before getting processed. - O_SYNC fd's operations now go through the queue to make sure previously acknowledged writes on the file (via other fds) are fulfilled before getting processed. - Option to not honor O_SYNC is now removed. - Option to ignore O_DIRECT is added (useful when running a VM and the drive appears with NCQ/TCQ or WCE=1 for the guest.) - Option to disable_first_nbytes is removed (as the cause of the bug which required this was diagnosed to be missing TCP_NODELAY.) - General cleanup and better conformance to coding style and convention. Change-Id: Ib44fb72da3727246b4a85174cb568c2f0231f6de BUG: 857673 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.org/3947 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Amar Tumballi <amarts@redhat.com>
* performance/io-cache: provide hook for dumping inode contextRaghavendra Bhat2012-09-191-29/+38
| | | | | | | | | Change-Id: Idcf24200ad8f0cf018ab92118f6b77cc4b9edcab BUG: 843787 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/3816 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* libglusterfs/dict: make 'dict_t' a opaque objectAmar Tumballi2012-09-061-4/+7
| | | | | | | | | | | | | | | * ie, don't dereference dict_t pointer, instead use APIs everywhere * other than dict_t only 'data_t' should be the valid export from dict.h * added 'dict_foreach_fnmatch()' API * changed dict_lookup() to use data_t, instead of data_pair_t Change-Id: I400bb0dd55519a7c5d2a107e67c8e7a7207228dc Signed-off-by: Amar Tumballi <amarts@redhat.com> BUG: 850917 Reviewed-on: http://review.gluster.org/3829 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: avoid deadlock while taking the statedump of fdsRaghavendra Bhat2012-08-191-6/+24
| | | | | | | | | | | * Provide a hook for forget Change-Id: Ide7ea6d4c6a7d0d93b81570cb544f2bbda526eeb BUG: 846916 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.org/3795 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: store the wb_inode in local before windingRaghavendra Bhat2012-08-111-1/+3
| | | | | | | | | | | | | | | * Store the write-behind's inode context in the local structure before winding the call so that in callback inode context is found. * Before returning EBADFD check if the inode context (wb_inode) is NULL, along with the inode type. Change-Id: If5a1c667efe6882a6efef1439cee3effc32ff9a7 BUG: 846536 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3796 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vbellur@redhat.com>
* core: reduce the usage of global variablesAmar Tumballi2012-08-031-1/+1
| | | | | | | | | | | | | | | | | | | | | | | * move all the 'logging' related global variables into ctx * make gf_fop_list a 'const' global array, hence no init(), no edits. * make sure ctx is allocated without any dependancy on memory-accounting infrastructure, so it can be the first one to get allocated * globals_init() should happen with ctx as argument not yet fixed below in this patchset: * anything with 'THIS' related globals * anything related to compat_errno related globals as its one time init'd and not changed later on. * statedump related globals Change-Id: Iab8fc30d4bfdbded6741d66ff1ed670fdc7b7ad2 Signed-off-by: Amar Tumballi <amarts@redhat.com> BUG: 764890 Reviewed-on: http://review.gluster.com/3767 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/io-cache: use pthread_mutex_trylock to hold mutex in statedumpsRaghavendra Bhat2012-08-021-7/+60
| | | | | | | | | | | | Do not use pthread_mutex_lock and gf_log functions while dumping information to statedump, to avoid deadlocks. Change-Id: I2bb4592126c5d1eacf614130815a839c4a90c59a BUG: 843787 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3758 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/quick-read: use pthread_mutex_trylock to hold mutex in statedumpsRaghavendra Bhat2012-08-021-3/+16
| | | | | | | | | | | | Do not use pthread_mutex_lock and gf_log functions while dumping information to statedump, to avoid deadlocks. Change-Id: I0d30b04ae322f131b7a42d277ebd4385e83769e6 BUG: 843789 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3764 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: maintain a per-inode request queue instead of ↵Raghavendra G2012-08-022-596/+769
| | | | | | | | | | | | | | | | maintaining per-fd path based operations like stat etc, whose results will be affected by writes have to be ordered with writes. With request queues maintained in inode this can be done naturally, than when they are maintained per open fd. Change-Id: Ibdde3b81366f642d07531632fc9062cb44fad2e7 BUG: 765443 Signed-off-by: Raghavendra G <raghavendra@gluster.com> Reviewed-on: http://review.gluster.com/712 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/read-ahead: use pthread_mutex_trylock to hold mutex in statedumpsRaghavendra Bhat2012-08-011-11/+18
| | | | | | | | | | Change-Id: I4491c96fc7d9879ba9040dcda3700c81d99ba184 BUG: 843796 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3766 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Pranith Kumar Karampuri <pranithk@gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/io-threads: dump the io-threads specific information in statedumpRaghavendra Bhat2012-07-302-2/+52
| | | | | | | | | Change-Id: Ib9b9eab2d2e90fce21ef32f0f884101938978859 BUG: 843377 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3736 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/io-threads: Provide option to turn off least-priorityPranith Kumar K2012-07-302-5/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | RCA: In cases when self-heal is in progress, self-heal fops are starved because of least-priority. This affects other fops with conflicting inode, entry locks with self-heal. Fix: This patch provides configuring enable/disable of least-priority. Additional changes: Moved RCHECKSUM fop to low instead of least because it will still affect the performance of other fops if RCHECKSUM is in LEAST priority. Tests: Tested that the enabling/disabling of fops is working fine. Tested that RCHECKSUM fop priority is assigned LOW when least-priority is disabled. Change-Id: I8d72872b33e2ac7e1ad3afb27582682b0eb98a80 BUG: 843704 Signed-off-by: Pranith Kumar K <pranithk@gluster.com> Reviewed-on: http://review.gluster.com/3743 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: detect short writes and pend an EIO errorBrian Foster2012-07-251-1/+15
| | | | | | | | | | | | | | | Write-behind returns write requests immediately and queues the request in memory for merging, etc. If a write is incomplete, pend an EIO error for the next fop. This ensures that write failures are not silent and potentially ignored. BUG: 809975 Change-Id: I0e0e6c8e710efab58ccfaf746501d00e459eb7ef Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3712 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/write-behind: preserve lk-owner while syncing writes.Raghavendra G2012-07-251-66/+143
| | | | | | | | | | | | | - This patch also makes syncing of non-overlapping but consecutive writes parallel. Till now only contiguous writes were synced parallely. Change-Id: Icf0d5ea373f30c79fcdc90ba44b7e7a1bc5f0111 BUG: 765141 Signed-off-by: Raghavendra G <raghavendra@gluster.com> Reviewed-on: http://review.gluster.com/269 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* remove useless if-before-free (and free-like) functionsJim Meyering2012-07-136-32/+12
| | | | | | | | | | | | 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>
* fuse/md-cache: add support for the 'fopen-keep-cache' mount optionBrian Foster2012-07-132-8/+39
| | | | | | | | | | | | | | | | | | | | | | fopen-keep-cache disables unconditional page-cache invalidations on file open in fuse (via FOPEN_KEEP_CACHE) and replaces that behavior with detection of remote changes and explicit invalidations from mount/fuse. This option improves local caching through the page cache and native client. This change defines a new 'invalidate' translator callback to identify when an inode's cache mapping has been determined to be invalid. md-cache implements the policy to detect and invoke inode invalidations. fuse-bridge and io-cache implement invalidate handlers to invalidate the respective caches (page cache in the case of fuse). BUG: 833564 Change-Id: I99818da5777eaf06276c1c0b194669f5bab92d48 Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-on: http://review.gluster.com/3584 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* md-cache: cache SELinux and Posix ACL xattrs only if enabledAnand Avati2012-07-041-5/+56
| | | | | | | | | | | | | | Fetch and cache SELinux and Posix ACL extended attributes only if they are enabled in the command line respectively. Fetching the extra extended attributes is pointless and negatively impacts performance Change-Id: I1bd1dbb1abb4a6929fad5f78bbfeaab8542ab4e2 BUG: 765785 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.com/3538 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com>
* localtime and ctime are not MT-SAFEKaleb S. KEITHLEY2012-06-292-8/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | There are a number of nit-level issues throughout the source with the use of localtime and ctime. While they apparently aren't causing too many problems, apart from the one in bz 828058, they ought to be fixed. Among the "real" problems that are fixed in this patch: 1) general localtime and ctime not MT-SAFE. There's a non-zero chance that another thread calling localtime (or ctime) will over-write the static data about to be used in another thread 2) localtime(& <64-bit-type>) or ctime(& <64-bit-type>) generally not a problem on 64-bit or little-endian 32-bit. But even though we probably have zero users on big-ending 32-bit platforms, it's still incorrect. 3) multiple nested calls passed as params. Last one wins, i.e. over- writes result of prior calls. 4) Inconsistent error handling. Most of these calls are for logging, tracing, or dumping. I submit that if an error somehow occurs in the call to localtime or ctime, the log/trace/dump still should still occur. 5) Appliances should all have their clocks set to UTC, and all log entries, traces, and dumps should use GMT. 6) fix strtok(), change to strtok_r() Other things this patch fixes/changes (that aren't bugs per se): 1) Change "%Y-%m-%d %H:%M:%S" and similar to their equivalent shorthand, e.g. "%F %T" 2) change sizeof(timestr) to sizeof timestr. sizeof is an operator, not a function. You don't use i +(32), why use sizeof(<var>). (And yes, you do use parens with sizeof(<type>).) 3) change 'char timestr[256]' to 'char timestr[32]' where appropriate. Per-thread stack is limited. Time strings are never longer than ~20 characters, so why waste 220+ bytes on the stack? Things this patch doesn't fix: 1) hodgepodge of %Y-%m-%d %H:%M:%S versus %Y/%m/%d-%H%M%S and other variations. It's not clear to me whether this ever matters, not to mention 3rd party log filtering tools may already rely on a particular format. Still it would be nice to have a single manifest constant and have every call to localtime/strftime consistently use the same format. Change-Id: I827cad7bf53e57b69c0173f67abe72884249c1a9 BUG: 832173 Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com> Reviewed-on: http://review.gluster.com/3568 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Anand Avati <avati@redhat.com>
* io-cache,quick-read: bring down log levelAnand Avati2012-06-072-4/+5
| | | | | | | | | | | | | | | | | | log messages were unnecessarily in INFO level. The two functions with the same name were non-static and actually the quick-read's call landed in the io-cache's version: 2012-06-07 17:02:29.848667] I [io-cache.c:1549:check_cache_size_ok] 0-single-master-io-cache: Max cache size is 33791991808 [2012-06-07 17:02:29.848751] I [io-cache.c:1549:check_cache_size_ok] 0-single-master-quick-read: Max cache size is 33791991808 Changed them to static declaration. Change-Id: Id9daf9593b2832e4c261f95eac6181efea8899a5 BUG: 765227 Signed-off-by: Anand Avati <avati@redhat.com> Reviewed-on: http://review.gluster.com/3536 Reviewed-by: Jeff Darcy <jdarcy@redhat.com> Tested-by: Gluster Build System <jenkins@build.gluster.com>
* performance/md-cache: make cache timeout reconfigurable via cliRaghavendra Bhat2012-05-291-3/+3
| | | | | | | | | | Change-Id: I73685bf73a14435884d3489343bbe71c331eb2a5 BUG: 825638 Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com> Reviewed-on: http://review.gluster.com/3458 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>
* performance/quick-read: fix race-conditions in qr_unlink.Raghavendra G2012-05-233-46/+80
| | | | | | | | | | | | | | | | | | The list of fds on which open needs to be done as part of unlink, was being modified at different places using different locks. This resulted in a race-condition where open was marked as in-transit, but fdctx was removed from the list of fds on which open was being sent even before open was done. Because of this, open_in_transit would be set forever (as an open was never actually sent, there would be no open_cbk called and hence we could not reset the variable), blocking all the future fd based fops on this fd. Change-Id: Ie84a55bee578869a9a060a094ba28480e7643ae8 BUG: 819490 Signed-off-by: Raghavendra G <raghavendra@gluster.com> Reviewed-on: http://review.gluster.com/3371 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vijay@gluster.com>
* performance/md-cache: use mdc_fsetattr_cbk as the callback if setattr is on a fdRaghavendra Bhat2012-05-211-1/+1
| | | | | | | | | | Change-Id: I114a16055540e0cd3317b83b329600251ffe03c3 BUG: 823886 Signed-off-by: Raghavendra Bhat <raghavendrabhat@gluster.com> Reviewed-on: http://review.gluster.com/3386 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Amar Tumballi <amarts@redhat.com> Reviewed-by: Anand Avati <avati@redhat.com>