statistics when several ganesha servers run on the same node
Hello, I meet an issue when collecting the statistics using `ganesha_stats` when several ganesha servers are started on the same node. Each of them use its own NFS port, path, export_id and pseudo. `ganesha_stats` returns the statistics of the first started ganesha server but fails to collect the statistics of the second one with `Export id not found`. The rpm in use is nfs-ganesha-utils-4.0-1.el8.x86_64. Is it a know problem and is it fix in a newer version ? Thanks for your help See below an example: - two ganesha servers are running on a single node. # ps -ef | grep ganesha root 2956841 1 0 09:04 ? 00:00:01 ganesha.nfsd -L /tmp/sbbtestgkhLzj8/ganesha/logfile -f /tmp/sbbtestgkhLzj8/ganesha/ganesha.conf -N NIV_EVENT -p /tmp/sbbtestgkhLzj8/ganesha/pidfile root 2957068 1 0 09:04 ? 00:00:00 ganesha.nfsd -L /tmp/sbbtestZoXOzPM/ganesha/logfile -f /tmp/sbbtestZoXOzPM/ganesha/ganesha.conf -N NIV_EVENT -p /tmp/sbbtestZoXOzPM/ganesha/pidfile # df | grep mypath localhost:/mypath.3714 104806400 35679232 69127168 35% /tmp/sbbtestgkhLzj8/mount localhost:/mypath.9019 104806400 35679232 69127168 35% /tmp/sbbtestZoXOzPM/mount - Write using each of them $ dd if=/dev/zero of=//tmp/sbbtestgkhLzj8/mount/fichier bs=1M count=10 10+0 records in 10+0 records out 10485760 bytes (10 MB, 10 MiB) copied, 0.0257701 s, 407 MB/s $ dd if=/dev/zero of=//tmp/sbbtestZoXOzPM/mount/fichier bs=1M count=10 10+0 records in 10+0 records out 10485760 bytes (10 MB, 10 MiB) copied, 0.0258967 s, 405 MB/s - Collect of stats for each of them # ganesha_stats iov42 3714 EXPORT 3714: requested transferred total errors latency READv42: 0 0 0 0 0 0 WRITEv42: 10485760 10485760 10 0 6141586 0 # ganesha_stats iov42 9019 EXPORT 9019: Export id not found # ganesha_stats export Export Stats Stats collected since: Thu Jun 22 09:04:20 2023814664066 nsecs Duration: 4897.5012726784 seconds Export id: 3714 Path: /tmp/sbbtestgkhLzj8/path NFSv3 stats available: 0 NFSv4.0 stats available: 0 NFSv4.1 stats available: 0 NFSv4.2 stats available: 1 MNT stats available: 0 NLMv4 stats available: 0 RQUOTA stats available: 0 9P stats available: 0 Export id: 0 Path: / NFSv3 stats available: 0 NFSv4.0 stats available: 0 NFSv4.1 stats available: 0 NFSv4.2 stats available: 1 MNT stats available: 0 NLMv4 stats available: 0 RQUOTA stats available: 0 9P stats available: 0
This is because the statistics uses DBUS, and the first ganesha binds the well-know DBUS address, and the rest cannot bind it. Daniel On 6/22/23 10:31, francoise.boudier--- via Support wrote:
Hello,
I meet an issue when collecting the statistics using `ganesha_stats` when several ganesha servers are started on the same node. Each of them use its own NFS port, path, export_id and pseudo.
`ganesha_stats` returns the statistics of the first started ganesha server but fails to collect the statistics of the second one with `Export id not found`.
The rpm in use is nfs-ganesha-utils-4.0-1.el8.x86_64. Is it a know problem and is it fix in a newer version ?
Thanks for your help
See below an example:
- two ganesha servers are running on a single node. # ps -ef | grep ganesha root 2956841 1 0 09:04 ? 00:00:01 ganesha.nfsd -L /tmp/sbbtestgkhLzj8/ganesha/logfile -f /tmp/sbbtestgkhLzj8/ganesha/ganesha.conf -N NIV_EVENT -p /tmp/sbbtestgkhLzj8/ganesha/pidfile root 2957068 1 0 09:04 ? 00:00:00 ganesha.nfsd -L /tmp/sbbtestZoXOzPM/ganesha/logfile -f /tmp/sbbtestZoXOzPM/ganesha/ganesha.conf -N NIV_EVENT -p /tmp/sbbtestZoXOzPM/ganesha/pidfile # df | grep mypath localhost:/mypath.3714 104806400 35679232 69127168 35% /tmp/sbbtestgkhLzj8/mount localhost:/mypath.9019 104806400 35679232 69127168 35% /tmp/sbbtestZoXOzPM/mount
- Write using each of them $ dd if=/dev/zero of=//tmp/sbbtestgkhLzj8/mount/fichier bs=1M count=10 10+0 records in 10+0 records out 10485760 bytes (10 MB, 10 MiB) copied, 0.0257701 s, 407 MB/s $ dd if=/dev/zero of=//tmp/sbbtestZoXOzPM/mount/fichier bs=1M count=10 10+0 records in 10+0 records out 10485760 bytes (10 MB, 10 MiB) copied, 0.0258967 s, 405 MB/s
- Collect of stats for each of them # ganesha_stats iov42 3714 EXPORT 3714: requested transferred total errors latency READv42: 0 0 0 0 0 0 WRITEv42: 10485760 10485760 10 0 6141586 0
# ganesha_stats iov42 9019
EXPORT 9019: Export id not found
# ganesha_stats export Export Stats Stats collected since: Thu Jun 22 09:04:20 2023814664066 nsecs Duration: 4897.5012726784 seconds Export id: 3714 Path: /tmp/sbbtestgkhLzj8/path NFSv3 stats available: 0 NFSv4.0 stats available: 0 NFSv4.1 stats available: 0 NFSv4.2 stats available: 1 MNT stats available: 0 NLMv4 stats available: 0 RQUOTA stats available: 0 9P stats available: 0
Export id: 0 Path: / NFSv3 stats available: 0 NFSv4.0 stats available: 0 NFSv4.1 stats available: 0 NFSv4.2 stats available: 1 MNT stats available: 0 NLMv4 stats available: 0 RQUOTA stats available: 0 9P stats available: 0
_______________________________________________ Support mailing list -- support@lists.nfs-ganesha.org To unsubscribe send an email to support-leave@lists.nfs-ganesha.org
Any plan to lift this restriction ? In the meantime, how can we get the statistics of several ganesha daemons on the same node ? Thanks a lot for your help, Francoise
participants (2)
-
Daniel Gryniewicz -
francoise.boudier@atos.net