This post is more than 5 years old
5 Practitioner
•
274.2K Posts
0
1719
Restart isi_gconfig_d
Hello
Can't we do IO during restarting isi_gconfig_d? If we can't do IO it's a problem, isn't it? Because a knowledge base article "isi_gconfig_d memory size caused node to panic with BUF_TIMELOCK(000202948)" recommends to restart it. However the situation of the KB is worse because the access is too slow?
Best Regards,
Hiroki
Stdekart
104 Posts
0
October 14th, 2015 11:00
Hiroki K.,
isi_gconfig_d is a process utilized in internal communication with lsass and lwio, restarting this processes will have no negative effects on client connections or access.
Please do keep in mind that while restarting the process fixes the issues it is temporary and likely to happen again.
As stated in the KB you referenced it is fixed in the following versions/patch of OneFS:
An upgrade should be planned at your earliest convenience to apply the permanent fix.
Shane
Anonymous User
170 Posts
1
October 14th, 2015 07:00
I've restarted isi_gconfig_d several times on busy clusters with no impact to my NFS clients.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
October 14th, 2015 17:00
Hello ed.wilts,
I didn't know that isi_gconfig_d works also for NFS. Because I found only the following on page 42 of "Isilon OneFS Release number 7.2.0.2.159228 DA OneFS Release Notes September 18, 2015" about what isi_gconfig_d does.
Lwio subscriptions held by the isi_gconfig_d process were not always released in a timely manner. As a result, the subscriptions sometimes accumulated. If a large number of subscriptions accumulated, it sometimes took a long time to release these resources back to the system and it was possible for the isi_gconfig_d process to become unresponsive until the operation was complete. Because the isi_gconfig_d process is responsible for maintaining SMB share configuration information, if this issue occurred, SMB clients were prevented from viewing or creating shares.
Thank you for answering.
Best regards,
Hiroki