[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Savannah-hackers] Re: subversions lossage
From: |
Loic Dachary |
Subject: |
[Savannah-hackers] Re: subversions lossage |
Date: |
Fri, 13 Jun 2003 12:40:37 +0200 |
Hi,
After a few minutes only, the ext2 partition shows:
/dev/hda3 2071416 551212 1414980 29% /tmpcvs
which suggests that the ram file system is no longer big
enough to hold the temporary data necessary to run pserver. The disk
I/O did not become histerical and the load average remained stable
(although high : ~6). However, after looking into the temporary
directories, it shows that some people are using something like:
cvs -d :pserver:address@hidden:/cvsroot co .
and this checks out *all* Savannah CVS trees (except private
ones). That "hole" was not too much of a problem until now and was
kept for backward compatibility purposes. At this point in time, I
guess it's not unreasonable to remove this possibily.
I moved /cvsroot/CVSROOT to /cvsroot/.CVSROOT-moved-by-loic-2003-06-13
If, after all, /tmpcvs is populated with less than 100MB in average,
we could switch back to using /mnt/ramfs. However, if it turns out that
using an ext2 partition solves the disk overload problem we had with ext3
partitions, I suggest we keep it that way. We could use some mount options
to further enhance the performances of the /tmpcvs partition, such as
-o async. But I'm not an expert on this matter an I guess rao knows better.
Cheers,
Loic Dachary writes:
>
> Hi,
>
> I'm going to switch from the ramfs file system (too small
> because too many simultaneous cvs pserver) to a dedicated ext2
> partition. I bet on the fact that the problem we experienced when
> using /tmp was linked to the specific behaviour of ext3 file systems.
>
> I'll let you know how it goes.
>
> --
> Loic Dachary http://www.dachary.org/ address@hidden
> 12 bd Magenta http://www.eucd.info/ address@hidden
> 75010 Paris T: 33 1 42 45 07 97 address@hidden
> GPG Public Key: http://www.dachary.org/loic/gpg.txt