[Fedora-directory-users] Re: Memory leak in FDS?
by Philip Kime
Hmm, I may have spoken too soon about the memory leak fix - it fixed it
when I tested doing about 4 requests per second over SSL but on the
server that had the real leak, the load is much higher, after an hour,
SNMP reports:
RHDS-MIB::dsOneLevelSearchOps.389 = Counter32: 155748
so it's getting a fair few requests and the memory is drifting up
slowly, by a megabyte or so every few minutes. I'll keep an eye on it
overnight but it looks like it's monotinically increasing ...
PK
--
Philip Kime
NOPS Systems Architect
310 401 0407
17 years, 7 months
[Fedora-directory-users] [FDS] Error: Cannot connect to the Admin Server
by Jorge Santos
Hi There,
I need a help about fds running.
i did the download of fedora-ds-1.0.2-1.FC4.i386.opt.rpm package and i
follow the instructions in
http://directory.fedora.redhat.com/wiki/Setup#Running_Setup
but, when i tried to execute the command:
# ServerRoot/startconsole -u admin -a http://localhost:1500
the console openned and after i put my password which was configured during
installation. But, the follow message is always returned:
Cannot connect to the Admin Server "http://localhost:1500" The url is not
correct or the server is not running.
I'd executed the follow tests:
Open my browser and put the url "http://localhost:1500" and the page
of Fedora Management Console was openned
execute the command telnet: telnet localhost 1500 and the follow
message was returned Connected to localhost.localdomain (127.0.0.1)
anybody have the some problem???
I'm using the FCS4
att,
Jorge Santos
17 years, 7 months
[Fedora-directory-users] Re: Memory leak in FDS?
by Philip Kime
I have performed some stress-testing of FDS 1.0.2 (64 bit) hitting the
server with about 4 requests/sec over SSL for ten hours and I can
reliably reproduce the memory leak. Under this load it leaks about 1Mb
ever 3-4 minutes. I took the server down, replaced all of the NSPR and
NSS libs in the /opt/fedora-ds tree with the new ones RM made available
nspr-4.6.2
nss-3.11.1
I then restarted the server and did exactly the same stress-test. No
leak at all in 14 hours under the same load. Looks fixed to me - nice
work, many thanks.
Here's the little script I used to replace all the libs quickly to
minimise downtime. Unpack the new libs to a folder and inside that
folder run this:
#!/bin/bash
for file in lib*
do
for efile in `find /opt/fedora-ds -name $file`
do
mv $efile $efile.bak
cp $file $efile
chmod 755 $efile
done
done
17 years, 7 months
[Fedora-directory-users] Building a core DS
by Eric Brown
Is there anyway to build an RPM that contains only the DS (without the
admin console, sample ldif's, etc.)? I saw something in the install
documentation about just running a core DS, but need to know if that
can be built.
17 years, 7 months
[Fedora-directory-users] Slapd stops periodically
by Timothy Sandel
I'm having an issue where slapd just stops periodically and there's no info in
the logs.
My setup:
fedora-ds-1.0.2-1.RHEL4 on RHAS4.4.
The userRoot database is setup as a consumer for an iPlanet 5.1 (win2k).
iPlanet is setup as single master and the replication agreement is set to
"always keep the directories in sync".
Works perfectly except every couple of days slapd on the fedora-ds box just
quits.
I googled the archives and don't see anything relevant.
Any pointers on how to debug would be appreciated.
Thanks,
Timothy
17 years, 7 months
[Fedora-directory-users] Replication of o=NetscapeRoot
by Brian Moyles
I've got 2 machines in multimaster replication across a WAN link. I'm
replicating our root suffix (userRoot) successfully. I'm storing
o=NetscapeRoot on box01 right now, and want to replicate that to 02 (using
2-way multimaster) and have 02 use its local copy so I have console failover
as described in the howto in the wiki.
What I'm unclear on, though, is where I should be creating the user for
replication. Right now, I have cn=Replication Manager, cn=config, meaning
that the user is in o=NetscapeRoot. The docs specify that the replication
user cannot exist in the database you're replicating...so where should it
go?
Thanks in advance!
Brian Moyles
Playboy Enterprises, Inc.
17 years, 7 months
[Fedora-directory-users] problem starting slapd
by Marco Bellacosa
Dear all,
I got problems while restarting my fedora-ds. In particular,
when I try to start the server via start-slapd I receive the following
message:
[23/Aug/2006:09:24:27 +0200] - SSL alert: CERT_VerifyCertificateNow:
verify certificate failed for cert server-cert of family
cn=RSA,cn=encryption,cn=config (Netscape Portable Runtime error -8181 -
Peer's Certificate has expired.)
[23/Aug/2006:09:24:27 +0200] - SSL failure: None of the cipher are valid\
Then, if I try to menage certificates via console, I am not able to log
in the console, I get the message:
Cannot connect to the Admin Server .....
The URL is not correct or the server is not running.
Therefore, I cannot start the server because my certificate is no more
valid and I cannot menage certificate because my console doesn't open
(it seems to me). Can anyone help me?
Thanks,
marco
17 years, 7 months
[Fedora-directory-users] account expiration
by Mikael Kermorgant
Hello,
I'd like to set up an expiration policy on my server but I've not
found any way of doing it in the docs.
I'd like to define for each user an expiration date.
Has someone on this list set up something similar ? If so, what
attribute have you used ?
Thanks in advance,
--
Mikael Kermorgant
17 years, 7 months
[Fedora-directory-users] Memory leak in FDS?
by Philip Kime
My new LDAP infra has started to be used fairly heavily recently and
today, it died - top reported:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
25708 ldap 15 0 11.9g 7.1g 3624 D 0.3 91.6 17168:00 ns-slapd
This looks like the memory leak I've seen rumours of? I'm using 64-bit
on CentOS, version 1.0.2.
PK
--
Philip Kime
NOPS Systems Architect
310 401 0407
17 years, 7 months