[Date Prev][Date Next]
I just ran an experiment where I set up a master and slave LDAP servers.
Both servers started with empty databases, and then I started an ldapadd
with a copy of an ldapsearch -L as the input. The input file is
186156004 bytes long.
The whole procedure cranked along pretty well, until I noticed that the
slurpd was growing and got to be 130 MB long. My system's data seg
ulimit was at 131072, and the program exited shortly thereafter, leaving
about 180 MB of replog files.
Also, attempting to run slurpd -o waits a few minutes and aborts after
a few moments with "realloc of 16384 bytes failed". Raising my dataseg
limit to 1GB, slurpd immediately jumps to 181MB and then it runs (which
explains the slight delay). So maybe it's not a memory leak, just
a hog as regards memory.
Another interesting message I get, now that I'm past the realloc failure, is:
Processing in one-shot mode:
61357 total replication records in file,
0 replication records to process.
Now it appears to be processing the 61357 records. Don't know why it says
0, but that's probably a simple problem to fix.