I'd asked if anyone knew why networker would sometimes run slow (original
question below). Apart from some comments that compressasm is worse on
gzip files (I knew that), the main useful comment was was that there is a
networker mailing list on majordomo@iphase.com ("subscribe networker")
Looking through the archives of this site, it appears it _might_ be a
Solaris 2.4 kernel bug. Hmm. I've upgraded the networker server to
101945-44 (I haven't done the client because a similar problem showed on
a Solaris 2.5 client!) and let's hope that works.
Thanks to all that responded (I'm ill and so a bit lazy to dig up the
names - apologies).
rgds
Stephen
Original question:
> Hi all. It's that dreaded program "networker" again....
>
> Every so often, networker goes real real slow. This mornings backup
> is a good example:
>
> Solaris savegroup: (notice) Servers completed, 4 clients (All Succeeded)
> Start time: Sun Jan 26 02:00:01 1997
> End time: Sun Jan 26 13:03:21 1997
>
> 11 hours??!! Yikes.
>
> Yet looking through the logs...
>
> apollo: /db level=full, 991 MB 00:23:44 5982 files
>
> 990Mb in 23 minutes. That's pretty good.
>
> And it carries on well...
>
> terra2: / level=full, 90 MB 00:08:49 13315 files
> terra2: /netra level=full, 112 MB 00:03:43 2178 files
> terra2: /usr/local level=full, 455 MB 00:17:51 19992 files
> terra2: /opt level=full, 379 MB 00:10:55 9457 files
> terra2: /export/home level=full, 475 MB 00:21:35 8485 files
>
> We've slowed down by this stage... 475Mb in 21 minutes. Almost half the
> speed. And next...
>
> terra2: /u level=full, 705 MB 04:13:34 767 files
>
> 4 hours! What???!!!
>
> The only difference I can think of between /u and /export/home is that
> /u has a lot of gzipped files on it (it's where I stick my source archives).
> apollo:/db is mainly Oracle dbs files (so full of holes). I'm using
> compressasm to compress backups. Is compressasm _that_ bad on gzip files?
> A factor of _eight_ slower??! Can't be....
>
> terra2 and the networker server are Sparc 5's Solaris 2.4, 32Mb RAM, connected
> via a 3Com 10Mb Etherswitch network. apollo is an Ultra 1, Solaris 2.5, on a
> different IP network connected via a Cisco AGS+ router and two 10Mb
> Etherswitches.
>
> If anyone can explain this, I'd be grateful.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:44 CDT