| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For distribute only volumes we can use the information for
local subvolumes to avoid syncop calls which goes through the
whole stack to fetch stat and entries.
A separate function gf_defrag_fix_layout_puredist is introduced.
TODO: A glusterd flag needs to be introduced in case we want to
fall back to run the old way.
Perf numbers:
DirSize - 1Million Old New %diff
Depth - 100 (Run 1) 353 74 +377%
Depth - 100 (Run 2) 348 72 +377~%
Depth - 50 246 122 +100%
Depth - 3 174 114 +52%
Change-Id: I67cc136cebd34092fd775e69f74c2d5b33d3156d
Fixes: #1242
Signed-off-by: Susant Palai <spalai@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem: When auth.allow list is set to some ip's, add-brick
operation is failing.
Cause: add-brick commands creates a temparary mount on the
bricks to set the extended attributes on the brick mount
points. When auth.allow list is set to default i.e, * (all)
we will not see any issue, but when it is set to certain ip's
add-brick operation fails as temparory mount on the bricks
fails because the peers are not part of auth.allow list.
Solution: When auth.allow list is already set, add all the
peers to the auth.allow list during add-brick operation.
the old list will be replaced in post commit phase.
As this can happen with replace-brick operation as well,
added code to handle it.
updates: #1391
Change-Id: I5ede8c35f05ab25ff431b88e074ddbe9c10a90f1
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem:
If a gfid is present in indices/xattrop folder while
the file/dir is actaully healthy and all the xattrs are healthy,
it causes lot of lookups by shd on an entry which does not need
to be healed.
This whole process eats up lot of CPU usage without doing meaningful
work.
Solution:
Set trusted.ec.dirty xattr of the entry so that actual heal process
happens and at the end of it, during unset of dirty, gfid enrty from
indices/xattrop will be removed.
Change-Id: Ib1b9377d8dda384bba49523e9ff6ba9f0699cc1b
Fixes: #1385
Signed-off-by: Ashish Pandey <aspandey@redhat.com>
|
|
|
|
|
|
|
| |
CID: 1430146
Change-Id: Icce4ffa0e78575b110e0cfd9d5cfd133141680c1
Updates: #1060
Signed-off-by: Sheetal Pamecha <spamecha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
On executiing the command gluster vol set help, an error
comes up in glusterd logs stating `undefined symbol: xlator_api`.
This issue is seen for the rpc-transport/socket.so file.
Fix:
The symbol `xlator_api` is not found in rpc-transport/socket.so
file as it is not a xlator but a shared object for transport.
In the `xlator.c` file, there is a function `xlator_volopt_dynload`,
which looks for the default values of the options available in gluster,
which is stored inside the respective xlator files for different voltypes.
In each of these files the `options` object is present which contains
the default values, which is therefore referenced from the `options` data
member of `xlator_api` object in case of xlators.But, since
`rpc-transport/socket.so` is not an xlator we don't have the `xlator_api`
object present to point to that object. So, in case of
`rpc-transport/socket.so` type we are accesing the `options` object
directly from the `xlator_volopt_dynload` function to fetch the default
values for the available options.
Fixes: #827
Change-Id: I3b2b0c1f2a11896be250aaca1a33a65b044991d5
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The setlk interrupt handler uses a 'fork' of the resolved
fuse state from setlk (a copy with some edits) to initiate
its own auxiliary fop. Thus the references stored in the
fuse states of the setlk fop and of its interrupt handler
are shared (apart from the ones edited by the interrupt
handler -- but the bulk of them remain as is). The lifetimes
of these references are tied to the setlk fop, which has
established them by properly claiming their backing
resources. To guarantee the validity of these references in
the interrupt context, we need to make sure that the setlk
fop did not reclaim the fuse state while the interrupt
handler is running.
In other words, the setlk fop needs to wait for the
termination of the interrupt handler, which is accomplished
by the 'sync' strategy of the interrupt API (passing
true for the 'sync' argument of
fuse_interrupt_finish_{fop,interrupt} functions).
Change-Id: I9a6dc76972507be4b7ba8d023cc876e5fddf813f
Updates: #1374
Signed-off-by: Csaba Henk <csaba@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
With 'sync' strategy, a fop's cbk waits for
the interrupt handler to finish by making a
call to fuse_interrupt_finish_fop() with
sync = true.
The wait is implemented by monitoring an
interrupt_state struct member via a condition
variable. However, due to broken code logic,
the pthread_cond_wait() call is never reached.
This change introduces a new member to the
fuse_interrupt_state_t enum (the type of
aforementioned struct member),
FUSE_INTERRUPT_WAITING_HANDLER, which is then
used for indicating the state of waiting for
the interrupt handler.
Change-Id: I72ab06c37f45ff8f212a6a632bac1f647af05cbd
Updates: #1374
Signed-off-by: Csaba Henk <csaba@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
NetBSD FUSE does not implement FUSE notification yet. This changes
makes this feature a configure time option so that it can be disabled.
Fixes: #1381
Change-Id: I3d977d8d69b57e1ac6957be84a9ddbb69b100893
Type: Bug
Signed-off-by: Emmanuel Dreyfus manu@netbsd.org
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In a cluster env: getspec() detects that volfile not found.
but further on, this return code is set by another call
so the error is lost and not handled.
As a result the server responds with ambiguous message:
{op_ret = -1, op_errno = 0..} - which cause the client to stuck.
Fix:
server side: don't override the failure error.
fixes: #1375
Change-Id: Id394954d4d0746570c1ee7d98969649c305c6b0d
Signed-off-by: Tamar Shacked <tshacked@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This directory holds the extended attribute backing store on FFSv1
filesystems. It should be ignored by glusterfs just like the .glusterfs
directory.
Fixes: #1361
Change-Id: Ia259be2af722a651473a312b5044b431c2b01b10
Type: Bug
Signed-off-by: Emmanuel Dreyfus <manu@netbsd.org>
|
|
|
|
|
|
|
|
|
|
| |
Replace the use of blacklist and whitelist within the code.
No functional changes.
Fixes: #1378
Change-Id: Iaa10e31c3d460f52bfd12da906789fccca5ba4d4
Signed-off-by: Yaniv Kaul <ykaul@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The scenario of setting an xattr to a dir, killing one of the bricks,
removing the xattr, bringing back the brick results in xattr
inconsistency - The downed brick will still have the xattr, but the rest
won't.
This patch add a mechanism that will remove the extra xattrs during
lookup.
This patch is a modification to a previous patch based on comments that
were made after merge:
https://review.gluster.org/#/c/glusterfs/+/24613/
fixes: #1324
Change-Id: Ifec0b7aea6cd40daa8b0319b881191cf83e031d1
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When EC successfully healed a directory it assumed that maybe other
entries inside that directory could have been created, which could
require additional heal cycles. For this reason, when the heal happened
as part of one index heal iteration, it triggered a new iteration.
The problem happened when the directory was healthy, so no new entries
were added, but its index entry was not removed for some reason. In
this case self-heal started and endless loop healing the same directory
continuously, cause high CPU utilization.
This patch improves detection of new files added to the heal index so
that a new index heal iteration is only triggered if there is new work
to do.
Change-Id: I2355742b85fbfa6de758bccc5d2e1a283c82b53f
Fixes: #1354
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
|
|
|
|
|
|
|
|
|
| |
When dirp is null, we should not call sys_closedir() on it.
fixes: #1379
Change-Id: I33633df983aeea11e9d685e41ed9ec58644b6258
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
[ 144s] afr-dir-read.h:15:1: warning: type of 'afr_opendir' does not match original declaration [-Wlto-type-mismatch]
[ 144s] 15 | afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd, dict_t *xdata)
[ 144s] | ^
[ 144s] afr-dir-read.c:71:1: note: type mismatch in parameter 5
[ 144s] 71 | afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd)
[ 144s] | ^
[ 144s] afr-dir-read.c:71:1: note: 'afr_opendir' was previously declared here
only a warning, more of a truth-and-beauty thing
Change-Id: I2d6ff3fa0a8c5e6ef36e090a6545eaf638752192
Updates: #1002
Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A rebalance process does a lookup for every file in the dir it is processing
before checking if it supposed to migrate the file.
In this issue there are two rebalance processses running on a replica subvol:
R1 is migrating the FILE.
R2 is not supposed to migrate the FILE, but it does a lookup and
finds a stale linkfile which is mostly due to a stale layout.
Then, it tries to unlink the stale linkfile and gets EBUSY
as the linkfile fd is open due R1 migration.
As a result a misleading error msg about FILE migration failure
due EBUSY is logged in R2 logfile.
Fix:
suppress the error in case it occured in a node that
is not supposed to migrate the file.
fixes: #1371
Change-Id: I37832b404e2b0cc40ac5caf45f14c32c891e71f3
Signed-off-by: Tamar Shacked <tshacked@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Fixing the unchecked return value issues reported by coverity scan
CID: 1400734
CID: 1400750
Change-Id: I3c953df9ade4a1548e41e18018edb1b041f7e15e
Signed-off-by: karthik-us <ksubrahm@redhat.com>
Updates: #1060
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Display which options were not changed from the default.
The user may have opted to change some global or volume options
from the default they were initially. Display '(DEFAULT)' if the values
used are those that were not explicitly set by the user.
Example output:
Option Value
------ -----
cluster.server-quorum-ratio 50
cluster.enable-shared-storage disable (DEFAULT)
cluster.op-version 80000
cluster.max-op-version 90000
cluster.brick-multiplex disable (DEFAULT)
cluster.max-bricks-per-process 250 (DEFAULT)
glusterd.vol_count_per_thread 100 (DEFAULT)
cluster.localtime-logging disable (DEFAULT)
cluster.daemon-log-level INFO (DEFAULT)
Since glusterfind uses the value, it is now filtering the value
and only picking the 1st word (which is the value itself) and ignores
the rest, which may now be '(DEFAULT)'.
Fixes: #1357
Change-Id: I7c59055158d099a5de38943f2169fd02c77f5d09
Signed-off-by: Yaniv Kaul <ykaul@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
'this' is used before it is defined,
therefore it can lead to a NULL dereference.
Fix:
Moved the definition of 'this', before it's use
to avoid NULL dereference.
Change-Id: I6ad382192129dfa3a206426e5610040e7a905be6
Updates: #1096
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
[ 202s] server-common.h:192:1: warning: type of 'server4_post_link' does not match original declaration [-Wlto-type-mismatch]
[ 202s] 192 | server4_post_link(server_state_t *state, gfx_common_3iatt_rsp *rsp, inode_t *inode, struct iatt *stbuf, struct iatt *pre, struct iatt *post);
[ 202s] | ^
[ 202s] server-common.c:847:1: note: type mismatch in parameter 7
[ 202s] 847 | server4_post_link(server_state_t *state, gfx_common_3iatt_rsp *rsp,inode_t *inode, struct iatt *stbuf, struct iatt *pre, struct iatt *post, dict_t *xdata)
[ 202s] | ^
[ 202s] server-common.c:847:1: note: 'server4_post_link' was previously declared here
goes back to glusterfs-6 at least.
Only a warning, more of a truth-and-beauty thing.
Change-Id: Ib4a74b2d00aec519e9d80a4bdeca20eb6273777f
Updates: #1002
Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Mark dir as missing in layout structure to be healed in
dht_selfheal_directory.
fixes: #1327
Change-Id: If2c69294bd8107c26624cfe220f008bc3b952a4e
Signed-off-by: Susant Palai <spalai@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Fixing use-after-free issues
Change-Id: I33e599da6b8159695e2c9e989421b5877d7200b5
CID: 1430117, 1430138, 1430116
updates: #1202
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Seen in fedora rawhide/33, and SUSE tumbleweed, in all versions
going back at least as far as glusterfs-6.
[ 351s] glusterd.c:68:13: warning: type of 'snap_mount_dir' does not match original declaration [-Wlto-type-mismatch]
[ 351s] 68 | extern char snap_mount_dir[PATH_MAX];
[ 351s] | ^
[ 351s] glusterd-snapshot.c:65:6: note: array types have different bounds
[ 351s] 65 | char snap_mount_dir[VALID_GLUSTERD_PATHMAX];
[ 351s] | ^
[ 351s] glusterd-snapshot.c:65:6: note: 'snap_mount_dir' was previously declared here
In this case it's only a warning, but certainly merits fixing.
Another case where a decl in a header file instead of open-coding
extern decls in multiple .c files would have been preferable.
Change-Id: Idc91e536a56a1a7717be83ed27698069e71dff67
Updates: #1002
Signed-off-by: Kaleb S. KEITHLEY <kkeithle@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch fixes a crash in FOPs that operate on really large sharded
files where number of participant shards could sometimes exceed
signed int32 max.
The patch also adds GF_ASSERTs to ensure that number of participating
shards is always greater than 0 for files that do have more than one
shard.
Change-Id: I354de58796f350eb1aa42fcdf8092ca2e69ccbb6
Fixes: #1348
Signed-off-by: Krutika Dhananjay <kdhananj@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
There has been either an explicit null
dereference or a dereference after null
check in some cases.
Fix:
Added the proper condition for null check
and fixed null derefencing.
CID: 1430106 : Dereference after null check
CID: 1430120 : Explicit null dereferenced
CID: 1430132 : Dereference after null check
CID: 1430134 : Dereference after null check
Change-Id: I7e795cf9f7146a633097c26a766f16b159881fa3
Updates: #1060
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added a check for NULL before dereferencing
the object as it may be NULL in few cases
inside the funtion. Also, added a check for
the negative value of gfid_idx.
CID: 1430140
CID: 1430145
Change-Id: Ib7d23459b48bbc471dbcccab6d20572261882d11
Updates: #1060
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Replace an over-engineered GF_SKIP_IRRELEVANT_ENTRIES() with
inline function gf_irrelevant_entry(), adjust related users.
Change-Id: I6f66c460f22a82dd9ebeeedc2c55fdbc10f4eec5
Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Fixes: #1350
|
|
|
|
|
|
|
|
|
| |
Reason for changing the log-level stated at the github isse
fixes: #1353
Change-Id: I21202075916c5a7525e5f26e7fb595efe7717b66
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
problem: while the rebalance is in progress, if a node is
rebooted rebalance v status shows the stats of this node as
0 once the node is back.
Reason: when the node is rebooted, once it is back
glusterd_volume_defrag_restart() starts the rebalance and
creates the rpc. but due to some race, rebalance process is
sending disconnect event, so rpc object is getting destroyed. As
the rpc object is null, request for fetching the latest stats is
not sent to rebalance process. and stats are shows as default values
which is 0.
Solution: When the rpc object null, we should create the rpc if the
rebalance process is up. so that request can be sent to rebalance
process using the rpc.
fixes: #1339
Change-Id: I1c7533fedd17dcaffc0f7a5a918c87356133a81c
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
"Program and Library Variables" section from Automake manual suggests
the following:
. . .
_LDADD and _LIBADD are inappropriate for passing program-specific linker
flags (except for -l, -L, -dlopen and -dlpreopen). Use the _LDFLAGS
variable for this purpose.
. . .
Therefore it is reasonable to move $(LIB_DL) additon from _LDFLAGS to
_LDADD and _LIBADD variables for program and library respectively.
Change-Id: Id8b4734c207ab28a08bcce683d316cdc7acb0bcd
Updates: #1000
Signed-off-by: Anoop C S <anoopcs@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Removed the macro 'GD_MSG_DICT_SERL_LENGTH_GET_FAIL'
from the glusterd-messages file as
'GD_MSG_DICT_ALLOC_AND_SERL_LENGTH_GET_FAIL' is
used in it's place
Change-Id: I69d7d95b5cb8f1bdd7e616d7a3e9539e891ba378
Fixes: #874
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Make sure -largp is used at link time
- PTHREAD_MUTEX_ADAPTIVE_NP is not available, use PTHREAD_MUTEX_DEFAULT instead
- Avoid non POSIX [[ ]] in scripts
- Do not check of lock.spinlock is NULL since it is not a pointer
(it is not a pointer on Linux either)
Change-Id: I5e04a7c552d24f8a473c2b837828d1bddfa7e128
Fixes: #1347
Type: Bug
Signed-off-by: Emmanuel Dreyfus <manu@netbsd.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
Some of the functions didn't had sufficient
logging of information in case of failure.
Fix:
Added log information in few functions in
case of failure indicating the cause of
such event.
Change-Id: I301cf3a1c8d2c94505c6ae0d83072b0241c36d84
fixes: #874
Signed-off-by: nik-redhat <nladha@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 620158475f462251c996901a8e24306ef6cb4c42.
The patch to revert is https://review.gluster.org/#/c/glusterfs/+/24613/
Reverting is required as comments were posted regarding a more
efficient implementation were made after the patch was merged.
A new patch will be posted to adress the comments will be posted.
updates: #1324
Change-Id: I59205baefe1cada033c736d41ce9c51b21727d3f
Signed-off-by: Barak Sason Rofman <redhat@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The scenario of setting an xattr to a dir, killing one of the bricks,
removing the xattr, bringing back the brick results in xattr
inconsistency - The downed brick will still have the xattr, but the rest
won't.
This patch add a mechanism that will remove the extra xattrs during
lookup.
fixes: #1324
Change-Id: Ibcc449bad6c7cb46bcae380e42e4496d733b453d
Signed-off-by: Barak Sason Rofman <bsasonro@redhat.com>
|
|
|
|
|
|
|
|
| |
Convert an ad-hoc hack to a regular library function gf_syncfs().
Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Change-Id: I3ed93e9f28f22c273df1466ba4a458eacb8df395
Fixes: #1329
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem: add-brick operation is failing when replica or disperse
count is not mentioned in the add-brick command.
Reason: with commit a113d93 we are checking brick order while
doing add-brick operation for replica and disperse volumes. If
replica count or disperse count is not mentioned in the command,
the dict get is failing and resulting add-brick operation failure.
fixes: #1306
Change-Id: Ie957540e303bfb5f2d69015661a60d7e72557353
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem:
On executing tests/features/flock_interrupt.t the following error log
appears
[2020-06-16 11:51:54.631072 +0000] E
[fuse-bridge.c:4791:fuse_setlk_interrupt_handler_cbk] 0-glusterfs-fuse:
interrupt record not found
This happens because fuse-interrupt-record is never sent on the wire by
getxattr fop and there is no guarantee that in the cbk it will be
available in case of failures.
Fix:
wind getxattr fop with fuse-interrupt-record as cookie and recover it
in the cbk
Fixes: #1310
Change-Id: I4cfff154321a449114fc26e9440db0f08e5c7daa
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem:
fuse on receiving interrupt for setlk sends clear-lock "fop"
using virtual-getxattr. At the moment blocked locks which are
cleared return EAGAIN errno as opposed to EINTR errno
Fix:
Return EINTR errno.
Updates: #1310
Change-Id: I47de0fcaec370b267f2f5f89deeb37e1b9c0ee9b
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also add a message saying this is to be used only
for 'debug' purpose only. This is helpful to corner the
issue to acl. There were recently many issues reported
related to permissions, and acl access denied bugs.
The bugs were elsewhere, but to validate them and to
get people back to service (in certain cases like oVirt,
where gluster volumes are used mostly by single user),
this option can be used.
Updates: #876
Change-Id: I7be4401153607e11c9efb831ab794df4176604df
Signed-off-by: Amar Tumballi <amar@kadalu.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
To keep consistency inside transactions started by locking an entry or
an inode, this change delays the removal of entries that are currently
locked by one or more clients. Once all locks are released, the removal
is processed.
It has also been improved the detection of stale inodes in the locking
code of EC.
Fixes: #990
Change-Id: Ic8ba23d9480f80c7f74e7a310bf8a15922320fd5
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Currently remove-brick commands follow sync-op framework. For code
extensibility (like, adding more phases in the trnasaction) we are
migrating the command to mgmt v3 framework.
fixes: #1164
Change-Id: I5d363223d6f9dc7a70b61adb9d3a5250e84a71b4
Signed-off-by: Sanju Rakonde <srakonde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem: See github issue for details.
Fix:
-In lookup if the entry exists in 2 out of 3 bricks, don't fail the
lookup with ENOENT just because there is an entrylk on the parent.
Consider quorum before deciding.
-If entry FOP does not succeed on quorum no. of bricks, do not perform
new entry mark.
Fixes: #1303
Change-Id: I56df8c89ad53b29fa450c7930a7b7ccec9f4a6c5
Signed-off-by: Ravishankar N <ravishankar@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Logs and other output carrying timestamps
will have now timezone offsets indicated, eg.:
[2020-03-12 07:01:05.584482 +0000] I [MSGID: 106143] [glusterd-pmap.c:388:pmap_registry_remove] 0-pmap: removing brick (null) on port 49153
To this end,
- gf_time_fmt() now inserts timezone offset via %z strftime(3) template.
- A new utility function has been added, gf_time_fmt_tv(), that
takes a struct timeval pointer (*tv) instead of a time_t value to
specify the time. If tv->tv_usec is negative,
gf_time_fmt_tv(... tv ...)
is equivalent to
gf_time_fmt(... tv->tv_sec ...)
Otherwise it also inserts tv->tv_usec to the formatted string.
- Building timestamps of usec precision has been converted to
gf_time_fmt_tv, which is necessary because the method of appending
a period and the usec value to the end of the timestamp does not work
if the timestamp has zone offset, but it's also beneficial in terms of
eliminating repetition.
- The buffer passed to gf_time_fmt/gf_time_fmt_tv has been unified to
be of GF_TIMESTR_SIZE size (256). We need slightly larger buffer space
to accommodate the zone offset and it's preferable to use a buffer
which is undisputedly large enough.
This change does *not* do the following:
- Retaining a method of timestamp creation without timezone offset.
As to my understanding we don't need such backward compatibility
as the code just emits timestamps to logs and other diagnostic
texts, and doesn't do any later processing on them that would rely
on their format. An exception to this, ie. a case where timestamp
is built for internal use, is graph.c:fill_uuid(). As far as I can
see, what matters in that case is the uniqueness of the produced
string, not the format.
- Implementing a single-token (space free) timestamp format.
While some timestamp formats used to be single-token, now all of
them will include a space preceding the offset indicator. Again,
I did not see a use case where this could be significant in terms
of representation.
- Moving the codebase to a single unified timestamp format and
dropping the fmt argument of gf_time_fmt/gf_time_fmt_tv.
While the gf_timefmt_FT format is almost ubiquitous, there are
a few cases where different formats are used. I'm not convinced
there is any reason to not use gf_timefmt_FT in those cases too,
but I did not want to make a decision in this regard.
Change-Id: I0af73ab5d490cca7ed8d07a2ce7ac22a6df2920a
Updates: #837
Signed-off-by: Csaba Henk <csaba@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Issue:
When a process has the open fd and the same file is
unlinked in middle of the operations, then file based
lookup fails with ENOENT or stale file
Solution:
When the file already open and fd is available, use fstat
to get the file attributes
Change-Id: I0e83aee9f11b616dcfe13769ebfcda6742e4e0f4
Fixes: #1281
Signed-off-by: Vinayakswami Hariharmath <vharihar@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
Add destroy calls for 'store_volinfo_lock' and 'lock' of volume info.
Move initialization of 'store_volinfo_lock' from glusterd_op_create_volume()
to common place, which is glusterd_volinfo_new() indeed.
Change-Id: I5fae4469f28eab80c4fa6f5947646528e6aedad7
Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Fixes: #1291
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Brick process are not properly attached on any cluster node while
some volume options are changed on peer node and glusterd is down on
that specific node.
Solution: At the time of restart glusterd it got a friend update request
from a peer node if peer node having some changes on volume.If the brick
process is started before received a friend update request in that case
brick_mux behavior is not workingproperly. All bricks are attached to
the same process even volumes options are not the same. To avoid the
issue introduce an atomic flag volpeerupdate and update the value while
glusterd has received a friend update request from peer for a specific
volume.If volpeerupdate flag is 1 volume is started by
glusterd_import_friend_volume synctask
Change-Id: I4c026f1e7807ded249153670e6967a2be8d22cb7
Credit: Sanju Rakaonde <srakonde@redhat.com>
fixes: #1290
Signed-off-by: Mohit Agrawal <moagrawal@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For example:
We have 3 nodes and create ec 3*(2+1) volume for
test-disperse-0/test-disperse-1/test-disperse-2 when we do
'gluster v heal test full' in node-1 that can in node-1/
node-2/node-3 glustershd's get op=GF_EVENT_TRANSLATOR_OP
and then do full heal in different disperse group.
Let us say we have 2X(2+1) disperse with each brick
from different machine m0, m1, m2, m3, m4, m5. and candidate_max is m5.
and do full heal so '*index' is 3 and !gf_uuid_compare(MY_UUID, brickinfo->uuid)
will be true in m3,and then m3's glustershd will be the heal-xlator.
Id: I5c6762e6cfb375aed32d3fc11fe5eae3ee41aab4
Signed-off-by: yinkui <13965432176@163.com>
Change-Id: Ic7ef3ddfd30b5f4714ba99b4e7b708c927d68764
fixes: bz#1724948
|
|
|
|
|
|
|
|
|
|
| |
The curr call to delete the xattr from the dict fails to find the key: dict_del_sizen(xdata, xattr_name);
This is beacuse keysize is calculated as sizeof of xattr_name which is a pointer, this lead to wrong size -> hash.
Fix: call to dict_deln which get keysize using strlen.
fixes: #1282
Change-Id: I23ce1f8f7928e9daa43bc3a9fa8d3611e81bbc36
Signed-off-by: Tamar Shacked <tshacked@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Problem:
AFR doesn't delay post-op for fsync fop. For fsync heavy workloads
this leads to un-necessary fxattrop/finodelk for every fsync leading
to bad performance.
Fix:
Have delayed post-op for fsync. Add special flag in xdata to indicate
that afr shouldn't delay post-op in cases where either the
process will terminate or graph-switch would happen. Otherwise it leads
to un-necessary heals when the graph-switch/process-termination
happens before delayed-post-op completes.
Fixes: #1253
Change-Id: I531940d13269a111c49e0510d49514dc169f4577
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
|