ZFS ZFS Errors while using ctld with FC card / iSCSI

luqasz

luqasz

Hi

I've recently bought two qle2460 4GB FC cards. One for freeBSD and one for linux box. I've compiled kernel to use target mode. All went fine. When I did simple dd (if=/dev/sdX of=/dev/null) on linux box I see ~200 MB/s. When I've started to use those block devices, I've noticed some strange behavior. Copying a file from local disk on linux machine to a FC exported block device, speed just sky rockets and then slowly decreases to a stand still. On freeBSD zpool iostat shows ~20MB/s constant write speed. Which is slow, because my disks can write at ~100MB/s. After some time I see errors in dmesg:

Code: 
ctlfe_onoffline: isp0 current WWNN 0x2000001b3214e3e8
ctlfe_onoffline: isp0 current WWPN 0x2100001b3214e3e8
isp0: Setting Channel 0 wwns to 0x2100001b329e76b6 0x2100001b3214e3e8
ctlfe_onoffline: isp0 new WWNN 0x2100001b329e76b6
ctlfe_onoffline: isp0 new WWPN 0x2100001b3214e3e8
ctlfe_onoffline: isp0 (path id 0) target role enable succeeded
ctlfeasync: WWPN 0x2100001b329e76b6 port 0x000001 path 0 target 0 arrived
(0:3:2/2): SYNCHRONIZE CACHE(10). CDB: 35 00 00 00 00 00 00 00 00 00
(0:3:2/2): Tag: 0x11e93c, type 1
(0:3:2/2): ctl_process_done: 315 seconds
isp0: CTIO7 completed with Invalid RX_ID 0x11e93c
isp0: isp_handle_platform_ctio: CTIO7[11e93c] seq 0 nc 1 sts 0x8 flg 0xa060 sns 0 resid 0 FIN
isp0: CTIO7 completed with Invalid RX_ID 0x11e93c
(0:3:2/2): SYNCHRONIZE CACHE(10). CDB: 35 00 00 00 00 00 00 00 00 00
(0:3:2/2): Tag: 0x120ca4, type 1
(0:3:2/2): ctl_process_done: 296 seconds
(0:3:2/2): SYNCHRONIZE CACHE(10). CDB: 35 00 00 00 00 00 00 00 00 00
(0:3:2/2): Tag: 0x123c6c, type 1
(0:3:2/2): ctl_process_done: 237 seconds
(0:3:2/2): SYNCHRONIZE CACHE(10). CDB: 35 00 00 00 00 00 00 00 00 00
(0:3:2/2): Tag: 0x126580, type 1
(0:3:2/2): ctl_process_done: 125 seconds
(0:3:2/2): SYNCHRONIZE CACHE(10). CDB: 35 00 00 00 00 00 00 00 00 00
(0:3:2/2): Tag: 0x126370, type 1
(0:3:2/2): ctl_process_done: 186 seconds

Those errors just keep popping up. Same problems using same LUNs on iSCSI but without errors in dmesg.
 
I don't see any follow-on, did you get that error figured out?

I have a TrueNAS Core sending the same signals. FreeBSD/TrueNAS is a dual port target, ESX hosts are dual port init, so 4 paths in round robin.
 
Hi, for my setup, this issue started to appear after upgrading to the latest version of firmware from Dell (HP has very old firmware) up to 3.29 and version 8.0.9. Then, applied a custom configuration suggested by Lenovo for VMware ESXi. This issue started after the update and changing the settings on every card. Currently is set Point-to-Point at 1 Gbps speed. Do you know why is happening this? My setup is on a Supermicro motherboard, TrueNAS 13, and an ISP2462 FC card (for TrueNAS), and an ISP2460 (for ESXi 6.7U3).
 
You must log in or register to reply here.
Back
Top