Hello all!

After some odd behaviour (can't allocate socket?). I had to restart my server last night,  Reboot in gui (I had to wait approx 15 mins to get the web page to open!) didn't work.

Powerdown didn't work (left it over an hour to do it's thing) so I had to drop power

 

The machine took a VERY long time to come back up and now it's up, things appear to be running but parity check is running VERY slowly at 340KBps and will take well over a month to complete. It's normal done in approx 20 hours.

 

I only see pre clear plugin issues in logs, where else should I be looking / what should I do next?

 

Thanks for any support!  Diag attached.

 

 

server-diagnostics-20170905-0627.zip

You onboard controller is configured as IDE, change to AHCI in the bios, you also appear to be using port multipliers, that's no good for performance, other than that nothing obvious.

I do have two small IDE drives but no multipliers :-O

which drives seem to be on multipliers (how do I identify these in the diag?)

 

Previous parity checks have all run quickly (no hardware changes)

2017-09-05, 22:10:41	21 hr, 25 min, 35 sec	Unavailable	Canceled	4
2017-09-03, 11:08:11	21 min, 17 sec	Unavailable	Canceled	0
2017-09-02, 18:53:16	20 hr, 53 min, 15 sec	26,6 MB/s	OK	0
2017-08-02, 12:30:19	14 hr, 30 min, 18 sec	38,3 MB/s	OK	0
2017-07-04, 02:41:18	14 hr, 37 sec	39,7 MB/s	OK	0
2017-07-02, 12:40:22	14 hr, 40 min, 21 sec	37,9 MB/s	OK	0
2017-06-29, 12:22:27	16 hr, 48 min, 46 sec	33,1 MB/s	OK	0
2017-06-02, 13:10:38	15 hr, 10 min, 37 sec	36,6 MB/s	OK	0
2017-05-29, 23:23:39	14 hr, 13 min, 12 sec	39,1 MB/s	OK	0
2017-05-22, 22:18:15	14 hr, 15 min, 37 sec	39,0 MB/s	OK	0

 

 

any tips / things I should check at all?

1 hour ago, fysmd said:

I do have two small IDE drives

 

Yes, but i wasn't talking about those, your onboard SATA controller is using IDE mode, it should be set to AHCI.

 

1 hour ago, fysmd said:

but no multipliers

 

the two 4 port Marvell  controllers you have:

 

Marvell Technology Group Ltd. 88SE9235 PCIe 2.0 x2 4-port SATA 6 Gb/s Controller

 

have 6 and 7 disks connected, if you're not using a port multiplier then it's probably included in the controller, can you link the controller used?

 

1 hour ago, fysmd said:

any tips / things I should check at all?

 

Reboot, change the onboard controller to AHCI and start another check to see if there is an improvement.

HI, more information:

 

I found that disk3 was causing issues, simply cat filename >/dev/null on a large files from each disk worked fine for all but disk3.  Files from disk three would produce no errors but nor were they reading data in any measurable way.  (my parity check was also seemingly hung like this).

 

I also found the AHCI setting and changed it, rebooted with disk3 still connected (I replaced the sata cable) and the machine hung while playing risers transaction logs (I waited all day!)

 

I've now disconnected drive3, started in safe mode and am waiting for other drives to mount:

Sep  9 10:52:05 server emhttp: shcmd (76): set -o pipefail ; mount -t reiserfs -o noatime,nodiratime /dev/md1 /mnt/disk1 |& logger
Sep  9 10:52:05 server kernel: REISERFS (device md1): found reiserfs format "3.6" with standard journal
Sep  9 10:52:05 server kernel: REISERFS (device md1): using ordered data mode
Sep  9 10:52:05 server kernel: reiserfs: using flush barriers
Sep  9 10:52:05 server kernel: REISERFS (device md1): journal params: device md1, size 8192, journal first block 18, max trans len 1024, max batch 900, max commit age 30, max trans age 30
Sep  9 10:52:05 server kernel: REISERFS (device md1): checking transaction log (md1)

 

My two controllers are these: https://www.amazon.co.uk/dp/B00ESFEI2E/ref=pe_385721_51767431_TE_dp_2

 

 

 

14 minutes ago, fysmd said:

My two controllers are these: https://www.amazon.co.uk/dp/B00ESFEI2E/ref=pe_385721_51767431_TE_dp_2

 

That controller has a built in port multiplier, they are not recommend both for performance reasons and for possible issues, it also uses a Marvell chipset, so another source of potential problems.

 

Still, performance should be much better than now, 30 to 40MB/s like your previous checks is what I would expect, current problem may or not be related to them, but if you have a different controller to test that would be the first thing to do.

 

 

I don't have a different controller to try

 

Wonder why it's suddenly happened..

 

So would a couple of these fellas http://www.supermicro.com/products/accessories/addon/aoc-sas2lp-mv8.cfm

be better?

 

I'm very confused, these seem to have good feedback in Limetech forums but are also Marvell chipset I think??

Those were good for unRAID v5, not v6, for v6 LSI controllers based on the SAS2008 chipset are recommended, e.g., 9201-8i, 9211-8i and clones (H310, M1015, etc)