search

junneyang / zumastor

Automatically exported from code.google.com/p/zumastor
0 stars 1 forks source link

zumastor writes are extremely slow when running on top of software raid #84

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
The problem was first reported by Will. When he ran zumastor on top of
software raid6, the zumastor volume appeared to be hanging to users because
writes were extremely slow and sync seemed to take forever to finish. We
have seen the problem on both xfs and ext3, so looks like it is file system
independent.

Original issue reported on code.google.com by jiahotc...@gmail.com on 26 Feb 2008 at 1:03

GoogleCodeExporter commented 9 years ago 
pgquiles probably needs this fixed before he can use in production, so 
targeting for 0.7

Original comment by daniel.r...@gmail.com on 4 Mar 2008 at 1:47

GoogleCodeExporter commented 9 years ago 
I'm using hardware RAID (HP SmartArray P400 controllers) and it worked fine 
until
recently.

Original comment by pgqui...@gmail.com on 4 Mar 2008 at 8:07

GoogleCodeExporter commented 9 years ago 
Wow, what a bad choice of words, let's rephrase it: I'm using hardware RAID and 
it
works fine for me (at least, it worked fine for me when I tested Zumastor last 
week
and discovered issue 85). It's issue 85 what's preventing me from using 
Zumastor in
production.

Original comment by pgqui...@gmail.com on 4 Mar 2008 at 8:16

GoogleCodeExporter commented 9 years ago 
Jiaying said 
"I have not seen the 'extremely extremely' slow 
(more than four hours to write 1G data) problem on my testing machine
again. That may be because of the fix of the bio.throttle patch. But the
writes are still quite slow (more than half an hour to write 1G data) both on
my single-spindle testing machine and on the x4500 box. The problem only
happened when zumastor has created a snapshot"
She also thinks it might be related to O_DIRECT.

But since pgquiles doesn't need this right now, let's
defer this issue.

Original comment by daniel.r...@gmail.com on 4 Mar 2008 at 8:54

GoogleCodeExporter commented 9 years ago 
It turned out this problem is not related to software raid. The x4500 box we 
were
trying just had really bad performance with O_DIRECT writes. So mark it as 
'wontfix'.

Original comment by jiahotc...@gmail.com on 26 Mar 2008 at 7:21

GoogleCodeExporter commented 9 years ago

Original comment by daniel.r...@gmail.com on 26 Mar 2008 at 9:17