1

we have two QNAP TS-1635 NASes (one per location), running at identical configuration and firmware versions. However, one of them has ever higher and higher load averages and actually needs to be restarted every week once load averages get too high so that our monitoring gives it a rest.

There is nothing in htop, iotop or iftop showing up that could explain the high values; also dstat doesn't show anything (here I only show an excerpt, nothing really changes):

# dstat -cdngylr
--total-cpu-usage-- -dsk/total- -net/total- ---paging-- ---system-- ---load-avg--- --io/total-
usr sys idl wai stl| read  writ| recv  send|  in   out | int   csw | 1m   5m  15m | read  writ
  0   1  99   0   0| 435k  558k|   0     0 | 105B  680B| 631   814 |2.36 2.13 2.01|20.6  15.9
  1   4  95   0   0|   0     0 | 209B 1160B|   0     0 | 479   721 |2.36 2.13 2.01|   0     0
  0   0 100   0   0|   0     0 | 104B  488B|   0     0 | 451   647 |2.36 2.13 2.01|   0     0
  0   0 100   0   0|   0     0 | 209B  488B|   0     0 | 433   590 |2.36 2.13 2.01|   0     0
  0   0 100   0   0|   0     0 | 104B  488B|   0     0 | 461   686 |2.36 2.13 2.01|   0     0
  1   1  99   0   0|   0     0 | 209B  488B|   0     0 | 455   653 |2.17 2.10 2.00|   0     0
  0   0 100   0   0|   0     0 | 104B  488B|   0     0 | 441   625 |2.17 2.10 2.00|   0     0
  0   0 100   0   0|   0     0 | 209B  488B|   0     0 | 457   666 |2.17 2.10 2.00|   0     0

What could explain the load averages going continuously up? It's not used for anything except afp/smb storage for TimeMachine. All disks are SMART OK, and no errors/warnings show up in dmesg.

Marco Schuster
  • 53
  • 7
  • The NAS got what role ? An iSCSI target, a share with backup, etc.. I ask to be sure it's not a connected gear that would cause the load – yagmoth555 Jan 23 '19 at 19:18
  • Have you made sure it’s not something as simple as a raid rebuild. These can take a long time dependent on the size and number of drives of your array – Timothy Frew Jan 24 '19 at 00:32
  • @yagmoth555 the NAS is a simple timemachine data dump for about 30 users, nothing fancy. The NAS on the other location (which does NOT make problems) has 100 timemachine users + serves as NFS for various virtual machines and other stuff, it's not the load that is the problem – Marco Schuster Jan 24 '19 at 12:59
  • @TimothyFrew I suspected this too, but RAID resyncs are a) only done once a week and b) these show up as disk IO load (it's all md-raid, no hardware raid) when they happen, but as seen in the dstat log there is no disk activity or network activity going on, despite the high load. – Marco Schuster Jan 24 '19 at 13:01

0 Answers0