illumos

2012-03-26

03:14 PM illumos gate Bug #2543 (New): bad duality in metaslab_df_alloc/metaslab_df_fragmented

in metaslab_df_alloc the rule to switch to first-fit is:
if (max_size < metaslab_df_alloc_threshold ||
... Emmanuel Hocdet

02:05 PM illumos gate Bug #2542 (New): dmu_prefetch: zfs_prefetch_disable should not disable metaslab_prefetch

activate zfs_prefetch_disable disable all dmu_prefetch.
For metaslab_prefetch is a very bad idea:
When zfs decide... Emmanuel Hocdet

2012-02-24

04:51 PM illumos gate Bug #2083: stmf: worker manager doesn't scale workers anymore

Dan McDonald wrote:
> Could you please test the one Alexander made? I ask because...
>
I tested it and it's ok.
Emmanuel Hocdet

2012-02-10

03:42 PM illumos gate Bug #2083: stmf: worker manager doesn't scale workers anymore

Dan McDonald wrote:
> Alexander Stetsenko wrote:
> > I guess it should look like this:
>
> Yeah, that's more lik... Emmanuel Hocdet

2012-02-08

11:07 AM illumos gate Bug #2083: stmf: worker manager doesn't scale workers anymore

> Emmanuel, could you provide more detailed description of the problem, what are the symptoms… ?
stmf_worker_mgmt(... Emmanuel Hocdet

2012-02-06

11:27 AM illumos gate Bug #2083 (Resolved): stmf: worker manager doesn't scale workers anymore

it's a regression introduced in changeset: 13291:d5c7e34a4e3f
(https://www.illumos.org/issues/434)
in the new fun... Emmanuel Hocdet

2010-12-09

04:30 PM illumos gate Bug #434: kernel crash dump on comstar iscsit

two crashs log on nexenta:
http://pastebin.com/1UT2guMN
two on openindiana
http://pastebin.com/hbpPRgzS
Emmanuel Hocdet

2010-11-30

03:46 PM illumos gate Bug #430: zfs create -b and -o volblocksize produce inconsistent results

yep different formula same result
50+50/65536*256 = ~50.2 Emmanuel Hocdet

01:54 PM illumos gate Bug #430: zfs create -b and -o volblocksize produce inconsistent results

50/512*(512+256) = 75
50/8192*(8192+256) = ~51.6
if you try 64k
50/65536*(65536+256) = ~50.2
-o volblock... Emmanuel Hocdet

03:33 PM illumos gate Bug #434 (Resolved): kernel crash dump on comstar iscsit

# for ((i=1; i<=64; i++)); do itadm create-target -n "iqn.1986-03.com.sun:02:$i"& done;
can crash kernel easily ... Emmanuel Hocdet