https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
Alan Somers changed:
What|Removed |Added
Status|New |Closed
Resolution|---
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #27 from David ---
so revisiting this, the issue is still here, tried to create a cluster cluster
with BSD 13.1 nodes and 2 BSD 13.1 clients and both clients can write to the
cluster volume the same file and the file will never
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #26 from Alan Somers ---
(In reply to David from comment #25)
As I explained before, you can do it by setting vfs.fusefs.data_cache_mode=0.
But that only works for fusefs daemons using protocol level 7.22 or below. For
higher
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #25 from David ---
ok so is there a way to disable all write cache in fuse?
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #24 from David ---
ok so is there a way to disable all write cache in fuse?
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #23 from Alan Somers ---
(In reply to David from comment #21)
I really can't help you. I'm an expert on fuse, not gluster. You need help
from Daniel, or perhaps from upstream. If you can provide a trace of fuse
operations, th
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #22 from David ---
another issue probably related to this:
after a few days a sever would drop gluster mounts due to memory exhaustion
pid 911 (glusterfsd), jid 0, uid 0, was killed: failed to reclaim memory
--
You are rece
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #21 from David ---
any updates on this ?
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #20 from David ---
pushing more data into the test file
[2022-11-04 07:25:27.171322] D [MSGID: 0]
[afr-transaction.c:2035:afr_post_nonblocking_lock_cbk] 0-codebase-replicate-0:
Non blocking locks done. Proceeding to FOP
[2
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #19 from David ---
started in foreground/debug mode
echo test > /codebase/test
[2022-11-04 07:17:28.063122] D [MSGID: 0]
[afr-transaction.c:2035:afr_post_nonblocking_lock_cbk] 0-codebase-replicate-0:
Non blocking locks done.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #18 from David ---
I have tried with:
fusefs rw,acl,transport=tcp,_netdev,direct-io-mode=disable
but the Warning message appears as soon as I modify a file on the mount
Will try and debug soon
--
You are receiving this mail
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #17 from David ---
I have tried with:
fusefs rw,acl,transport=tcp,_netdev,direct-io-mode=disable
but the Warning message appears as soon as I modify a file on the mount
Will try and debug soon
--
You are receiving this mail
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #16 from Alan Somers ---
(In reply to David from comment #15)
That option is actually parsed by the daemon. Most daemons call it
"direct_io", but maybe Gluster doesn't. Check Gluster's documentation.
--
You are receiving thi
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #15 from David ---
that's the first thing I tried but always got: Invalid option direct_io
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #14 from Alan Somers ---
(In reply to David from comment #13)
That "-o direct_io" would be if you're invoking the /sbin/mount_fusefs binary.
To put it in fstab, just add a ,direct_io along with the rest of the options.
--
You
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #13 from David ---
ok so I caught an error on the console in the lines of:
WARNING: FUSE protocol violation for server mounted at /codebase: cache
incoherent! To prevent data corruption, disable the data cache by mounting with
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #12 from Alan Somers ---
(In reply to David from comment #10)
That's the Gluster release version, not the fuse protocol version. The easiest
way to find out the fuse protocol version that it's using would be to run it in
verbos
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #11 from David ---
```
[root@au-syd01-qa-app02 ~]# sysctl -a | grep vfs.fusefs
vfs.fusefs.kernelabi_minor: 29
vfs.fusefs.kernelabi_major: 7
vfs.fusefs.stats.lookup_cache_misses: 685
vfs.fusefs.stats.lookup_cache_hits: 4724
vfs.f
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #10 from David ---
gluster is 8.4_2
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #9 from Alan Somers ---
(In reply to David from comment #8)
> it also looks like vfs.fuse.lookup_cache_enable is no longer an option as
> well > as I was trying to see if that would make a difference
That's because it's named
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #8 from David ---
I have tried to turn it off
```
[root@au-syd01-qa-app02 ~]# sysctl vfs.fuse.data_cache_mode = 0
sysctl: unknown oid 'vfs.fuse.data_cache_mode'
sysctl: unknown oid ''
sysctl: unknown oid '0'
[root@au-syd01-qa-a
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #7 from David ---
sorry the direct-io-mode=disable option
because I'm experiencing the issue with direct-io-mode=disable as well, it
seems to be cache related because deleting/creating files is instant and works
as expected
we
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #6 from Alan Somers ---
(In reply to David from comment #5)
> Hmmm FUSE direct_io mount option which I have used or sysctl
Yes, fusefs does implement direct_io.
> does fuse evict invalidated cache contents on fuse_write_direct
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #5 from David ---
root@au-syd01-qa-app01:~ # sysctl -a |grep vfs.fusefs.data_cache_mode
vfs.fusefs.data_cache_mode: 1
root@au-syd01-qa-app01:~ #
Hmmm FUSE direct_io mount option which I have used or sysctl
vfs.fuse.data_cache_
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #4 from David ---
root@au-syd01-qa-app01:~ # sysctl -a |grep vfs.fusefs.data_cache_mode
vfs.fusefs.data_cache_mode: 1
root@au-syd01-qa-app01:~ #
Hmmm FUSE direct_io mount option which I have used or sysctl
vfs.fuse.data_cache_
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
Alan Somers changed:
What|Removed |Added
CC||asom...@freebsd.org
--- Comment #3 f
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
--- Comment #2 from David ---
I have tried disabling direct IO, no change
--
You are receiving this mail because:
You are the assignee for the bug.
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=266879
Graham Perrin changed:
What|Removed |Added
Product|Ports & Packages|Base System
Summary|Glu
28 matches
Mail list logo
