[lustre-discuss] changelog not purging
jwhite at lbl.gov
Thu Apr 16 17:01:41 PDT 2020
Running 2.12.3. It doesn’t really look like that Jira issue.
So I just remounted and it forced a changelog cleanup (a deregister was interrupted by an MDS crash). So that cleaned up some 20% of the total capacity of the MDT (good news) but looking at the lctl outputs, the changelog_size is growing with no changelogs defined:
[root at io011 ~]# lctl get_param mdd.brc-MDT0000.changelog_users; lctl get_param mdd.brc-MDT0000.changelog_size; sleep 10; lctl get_param mdd.brc-MDT0000.changelog_size
current index: 2032979833
ID index (idle seconds)
[root at io011 ~]#
Also, even now cleaned, the file system with 563,353,202 inodes is using 930GB, that strikes me as a bit strange.
So even with the changelog users cleared, I suspect I’ll be back to capacity issues in another 7 days or so.
> On Apr 16, 2020, at 2:12 PM, Cameron Harr <harr1 at llnl.gov> wrote:
> What Lustre version are you on? Does it look like this issue: https://jira.whamcloud.com/browse/LU-12098
> On 4/16/20 1:35 PM, John White wrote:
>> We’re having some pretty intense MDT capacity growth issues on one of our file systems that coincide exactly with changelogs. Problem is we can’t seem to purge them. We can deregister them but the capacity remains. For a 750mil files we have close to 3TB of capacity usage on the MDT.
>> I know there are 2 files in the root of the ldiskfs for the changelog but those are only a couple MB max. Are there other files in the ldiskfs directory structure that can be manually purged?
>> lustre-discuss mailing list
>> lustre-discuss at lists.lustre.org
> lustre-discuss mailing list
> lustre-discuss at lists.lustre.org
More information about the lustre-discuss