Struggling with an interesting issue, I changed our network migrating to 10g on my main video editing box along with a new switch and SFP+ cables. My 2 Truenas units are HPE DL380 G8 units that until now were on a Gig-E network with a 2nd 10G nic using a 12" SFP+ cable between them. My primary network is 10.1.0.x/21 and the 2 NAS units have a 10g dual port HP card on the 172.16.10.x/29 network as peer to peer. Replications, file serving, uploads, and downloads from either NAS box have not had any issues until now.
I spent about 3 hours running iPerf tests tuning the throughput from the 10g nic in my editing system through the switch and up to the truenas-01 to get the best results. I can pull a 6g video from the NAS or push one to the NAS with no issue although there was a bit of a hiccup observed during one of the download tests that I caught on my ILO screen as:
FYI - ql1 is the 10g nic in the server that is connected to the switch assigned 10.1.0.91/21
qla_dmamap_callback: bus_dmamap_load failed (27)
ql1: qla_get_mbuf: bus_dmamap_load failed
ql1: qla_replenish_jumbo_rx: qla_get_mbuf [1,(1024),(2036)] failed
FYI - ql0 is the 2nd port of that card and is assigned as 172.16.10.1/29 for the peer to peer to nas-02
A test transfer of a 6gb file seemed to hiccup but continued and finished with no issue.
While transferring another file a 2nd much more severe issue occurred and was able to be duplicated a number of times which I'm sure is not good for my pools with random reboots and whatnot. This was odd in the sense that I pulled a file down from nas-01 to move the file to a mapped share on my editing box that is mapped back to nas-01. So basically go to my N Drive which is \\10.1.0.91\bigmovie\Holding Pen\ <project folder> and move it to my V drive a temp directory which is V:\_ Customer Folders or \\10.1.0.91\ml-desktop-bu\_ Customer Folders which each time resulted in a nasty sudden crash and reboot.
I was able to capture this from the screen as logs were of no use, any idea what these might be?
Thanks,
Miguel
I spent about 3 hours running iPerf tests tuning the throughput from the 10g nic in my editing system through the switch and up to the truenas-01 to get the best results. I can pull a 6g video from the NAS or push one to the NAS with no issue although there was a bit of a hiccup observed during one of the download tests that I caught on my ILO screen as:
FYI - ql1 is the 10g nic in the server that is connected to the switch assigned 10.1.0.91/21
qla_dmamap_callback: bus_dmamap_load failed (27)
ql1: qla_get_mbuf: bus_dmamap_load failed
ql1: qla_replenish_jumbo_rx: qla_get_mbuf [1,(1024),(2036)] failed
FYI - ql0 is the 2nd port of that card and is assigned as 172.16.10.1/29 for the peer to peer to nas-02
A test transfer of a 6gb file seemed to hiccup but continued and finished with no issue.
While transferring another file a 2nd much more severe issue occurred and was able to be duplicated a number of times which I'm sure is not good for my pools with random reboots and whatnot. This was odd in the sense that I pulled a file down from nas-01 to move the file to a mapped share on my editing box that is mapped back to nas-01. So basically go to my N Drive which is \\10.1.0.91\bigmovie\Holding Pen\ <project folder> and move it to my V drive a temp directory which is V:\_ Customer Folders or \\10.1.0.91\ml-desktop-bu\_ Customer Folders which each time resulted in a nasty sudden crash and reboot.
I was able to capture this from the screen as logs were of no use, any idea what these might be?
Thanks,
Miguel
