There still seems to be memory-leak-problems whith the V3-clients..?


With the 3.1.0.5-cliients, the memory-leak-problem was reported as fixed
(IC20230)

I don't know the details on this problem, as I had 3.1.0.6 as the
"entry-level" when I started testing the V3-clients.

I have a few digital-unix, solaris, aix and hp-clients running both
3.1.0.6 and 3.1.0.7 now, and the all seem to still suffer from this leak..

In terms of files processed/backed up, the larger clients suffer more than
the smaller.

In the worst cases the scheduler won't run for more than a week...

-----
One digital-unix-client (huge # of files) which in V2 was scheduled whith
One digital-unix-client (huge # of files) which in V2 was scheduled whith
the -slowincr-option is no longer able to complete with the
-memoryeff-option. This client has 512MB of memory, but the scheduler
takes it all before it dies....
takes it all before it dies....

On another (Solaris)-client which I hace to restart weekly, our
solaris-sysadm took a look whith the /usr/proc/bin/pmap <pid>, and this
shows that all the accumulated memory is of the type "heap"...
I'm no solaris expert, and the sysadm couln't explain it further either..

The rest of our about 600 V2-clients will have to wait until this has been
sorted out..

-----
One funny thing though......
One funny thing though......

I have about 20 of the  unsupported linux V3-clients running.
One of them, quite large, has an uptime today of 137 days. And it still
uses no more memory than its initial 1.8MB..

(Which probably prooves that the problem was introduced somewhere
inbetween 3.1.0.2 and 3.1.0.4)...


I guess I'm not the only one having the problem.
How do the rest of you cope with this?


                                         Tom