Lustre devel - Mar 2007 - [Bug 11980] Please watch the 1.6.0 builds

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



Request 1500 for v1_6_0_RC1 has been submitted.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



I have submitted request 1506 for v1_6_0_RC2.  Please watch it.  Note that this
is a lower priority than the 1.4.10 beta.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



We''re now onto RC3.  I have canceled all my previous requests and
submitted
request 1511 for this.  Please watch it as your highest priority.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



(In reply to comment #13)
> [2] single_mount - yujian found that there isn''t ldiskfs module
even in the
> lustre rpm package
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
>     https://buffalo.lustre.org:8443/display_report.pl?report_id=344326
2.4 does not use ldiskfs, so that''s not the problem.  (The message
you''re seeing
is just a hint as to something that _might_ be wrong, not a definite indication
of what the problem is.)  Looks like there''s an issue with the block
device:

Unable to mount /dev/disk/unique-id/SFUJITSU_MAP3735NP_UPJ0P46008MC: No such
device

So either the test is trying to use an incorrect device, or it''s
broken.  Please
fix and recycle.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



(In reply to comment #19)
> We have to modify ltest codes to explicitly pass
"--backfstype=ext3" (for 2.4)
> or "--backfstype=ldiskfs" (for 2.6) to tunefs.lustre. 
Can this be done soon?  We really need to move 1.6.0 forward.

Also, out of curiosity, why has this never been a problem before?  Has the test
system recently changed?

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



(In reply to comment #22)
> Have we ever tested mountconf with 2.4 kernels before?  I thought 1.6 was
going
> to mark the end of support for 2.4 kernels.
       
Yes.  There have been many 1.6.0 builds so far that have been done on RHEL 3.

In 1.6.0, we support RHEL 3 for clients only.  Unfortunately, the way ltest is
currently setup requires us to build a RHEL 3 server to allow the RHEL 3 clients
to be tested.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



(In reply to comment #32)
> [5] write_disjoint - failed due to timeout
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
>     https://buffalo.lustre.org:8443/display_report.pl?report_id=345914 
> 
> [6] replay-dual - OOPS found in
> /testsuite/ltest-boulder/harness/replay-dual-console-client3.log
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
>     https://buffalo.lustre.org:8443/display_report.pl?report_id=345807
> 
> 
> [7] replay-single - found lustre error
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
>     https://buffalo.lustre.org:8443/display_report.pl?report_id=345808
> 
> [8] sanity-norestarts - different stripe size and count
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
>     https://buffalo.lustre.org:8443/display_report.pl?report_id=345843
Shouldn''t bugs be logged on these issues?  For now, I have asked Beaver
to
investigate them as before.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980


> > [8] sanity-norestarts - different stripe size and count
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
> >     https://buffalo.lustre.org:8443/display_report.pl?report_id=345843
> 
for resolve this issue selinux need to be disabled. 
SLES10 has extra checks in setacl and allow only root changed not trusted ACL.

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11980



(In reply to comment #36)
> Probably it makes sence to increase timeout. Is this expected that this
test in
> 2.6-suse can be twice slower that it is in 2.6-sles10?      
It''s like that on b1_4 and has been for quite some time, so
I''m not concerned.
I think the timeout needs to be increased.