Hello!
I am having trouble with the NFS server dying under load when I am scanning my library in XBMC. XBMC starts scanning, and all the folder are exported in FreeNAS, and thumbnails are written to the shared NFS folder. After a short while, the scan stops, and in the XBMC logs just states that the folder is no longer availiable. A couple of minutes later, I am able to start the scan again, and it will continue for a short while before it dies again.
My FreeNAS server is running as a virtual machine on ESXi 5.5. The ESXi server has the following specs:
2x Intel Xeon X5660 (6x2,8GHz)
80GB DDR3 ECC
Supermicro X8DT3-LN4-F
4x 750GB in RAID 10 on a Dell PERC 6 w/ BBU for datastore
1x LSI 9211-8i (passthrough to FreeNAS)
1x LSI 1068E (passthrough to FreeNAS)
1x Intel onboard AHCI ((passthrough to FreeNAS)
FreeNAS setup:
4 cores dedicated to FreeNAS (no sharing with other VMs)
36GB RAM
Disks:
6x 3TB in RAIDZ2
6x 2TB in RAIDZ2
6x 1TB in RAIDZ2
2x Intel 320 40GB as ZIL mirror (not used, will probably remove them)
The storage pool is encrypted, if that matters.
When updating the library using SMB, it works fine, and the connection is not broken, even if the thumbnails are written to the NFS share. So it appears that it dies due to the combination of exporting many folders (836 movies) and writing all the thumbnails at the same time. I have attached the debug logs, and screenshots of the shares.
I had the same setup on 9.2, and I do believe that this did not happen then, although I could be wrong.
I have tried to add the IP/hostnames of all my XBMC clients in the host name database, but this did not help. I also dedicated the CPUs to the FreeNAS VM to avoid congestion of the CPU resources, but this did not help.
Please advise.
I am having trouble with the NFS server dying under load when I am scanning my library in XBMC. XBMC starts scanning, and all the folder are exported in FreeNAS, and thumbnails are written to the shared NFS folder. After a short while, the scan stops, and in the XBMC logs just states that the folder is no longer availiable. A couple of minutes later, I am able to start the scan again, and it will continue for a short while before it dies again.
My FreeNAS server is running as a virtual machine on ESXi 5.5. The ESXi server has the following specs:
2x Intel Xeon X5660 (6x2,8GHz)
80GB DDR3 ECC
Supermicro X8DT3-LN4-F
4x 750GB in RAID 10 on a Dell PERC 6 w/ BBU for datastore
1x LSI 9211-8i (passthrough to FreeNAS)
1x LSI 1068E (passthrough to FreeNAS)
1x Intel onboard AHCI ((passthrough to FreeNAS)
FreeNAS setup:
4 cores dedicated to FreeNAS (no sharing with other VMs)
36GB RAM
Disks:
6x 3TB in RAIDZ2
6x 2TB in RAIDZ2
6x 1TB in RAIDZ2
2x Intel 320 40GB as ZIL mirror (not used, will probably remove them)
The storage pool is encrypted, if that matters.
When updating the library using SMB, it works fine, and the connection is not broken, even if the thumbnails are written to the NFS share. So it appears that it dies due to the combination of exporting many folders (836 movies) and writing all the thumbnails at the same time. I have attached the debug logs, and screenshots of the shares.
I had the same setup on 9.2, and I do believe that this did not happen then, although I could be wrong.
I have tried to add the IP/hostnames of all my XBMC clients in the host name database, but this did not help. I also dedicated the CPUs to the FreeNAS VM to avoid congestion of the CPU resources, but this did not help.
Please advise.
Attachments
-
debug-midgard-20150124215036.tgz622.9 KB · Views: 314
-
2015-01-24 22_21_36-midgard - FreeNAS-9.3-STABLE-201501212031.png21.4 KB · Views: 295
-
2015-01-24 22_21_54-midgard - FreeNAS-9.3-STABLE-201501212031.png20.8 KB · Views: 282
-
2015-01-24 22_22_10-midgard - FreeNAS-9.3-STABLE-201501212031.png16.9 KB · Views: 283