ryan80
asked on
troubleshooting LDAP on fedora
I have recently been tasked with troubleshooting some issues with some Fedora 14 servers. There are several servers and they use LDAP for authentication which is located on one of the Fedora servers.
however the mounts on the servers will become unavailable every 1-2 weeks. At that time the LDAP server has to be restarted and then the mounts are available again. Right now these servers are not critical, but we are going to be relying on them more and more.
Now I have used Fedora a little bit and have a very basic understanding of it. I am primarily a Windows system admin. But I need to start trouble shooting these issues.
What I need is a place to start to find out what the issues are. I am sure that there are log files somewhere that will point me in the correct direction, but I have not the slightest on where to start looking.
Does anyone have any clue where to start looking, or where I can start looking to get more information for you?
however the mounts on the servers will become unavailable every 1-2 weeks. At that time the LDAP server has to be restarted and then the mounts are available again. Right now these servers are not critical, but we are going to be relying on them more and more.
Now I have used Fedora a little bit and have a very basic understanding of it. I am primarily a Windows system admin. But I need to start trouble shooting these issues.
What I need is a place to start to find out what the issues are. I am sure that there are log files somewhere that will point me in the correct direction, but I have not the slightest on where to start looking.
Does anyone have any clue where to start looking, or where I can start looking to get more information for you?
ASKER
Thanks for fast reply.
1. I believe that it is directory server. I am not 100% sure how to tell, but I went into /etc and looked into the openldap folder and it looked to be default. I went into the dirsvr folder and there are some folder that match the server name. I also went into the log folder for dirsrv and found some entries in the error log.
Not listening for new connections - too many fds open
2. I believe that the mounts in question are NFS shares from the LDAP server. they use autofs to mount them on demand.
3. The mounts in question are NFS
4. I on the LDAP server there is the /boot which is on /dev/sda1, the root is on a logvol00 which is on sda2, and the moutns in question are from a SAN /dev/mapper/mpath0p1. this last one mapper/mpath0p1, this is the connection to the SAN i assume and this is a multipathing device for the Fibre Channel?
1. I believe that it is directory server. I am not 100% sure how to tell, but I went into /etc and looked into the openldap folder and it looked to be default. I went into the dirsvr folder and there are some folder that match the server name. I also went into the log folder for dirsrv and found some entries in the error log.
Not listening for new connections - too many fds open
2. I believe that the mounts in question are NFS shares from the LDAP server. they use autofs to mount them on demand.
3. The mounts in question are NFS
4. I on the LDAP server there is the /boot which is on /dev/sda1, the root is on a logvol00 which is on sda2, and the moutns in question are from a SAN /dev/mapper/mpath0p1. this last one mapper/mpath0p1, this is the connection to the SAN i assume and this is a multipathing device for the Fibre Channel?
I am sure that if you run ulimit -a command then you will get the file discriptors as 1024 so please increase them as per the following.
I have faced the same issue.
echo "fs.file-max = 64000" >> /etc/sysctl.conf
echo "* soft nofile 8192" >> /etc/security/limits.conf
echo "* hard nofile 8192" >> /etc/security/limits.conf
echo "ulimit -n 8192" >> /etc/profile
sysctl -p
source /etc/profile
I have faced the same issue.
echo "fs.file-max = 64000" >> /etc/sysctl.conf
echo "* soft nofile 8192" >> /etc/security/limits.conf
echo "* hard nofile 8192" >> /etc/security/limits.conf
echo "ulimit -n 8192" >> /etc/profile
sysctl -p
source /etc/profile
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
thanks, this looks like a great place to start. one last question,
the last 2 commands that you have there, sysctl -p and source.
when I run sysctl -p, will those setting take effect immediately?
Also when I run source /etc/profile, I see that there are several scripts that are in there. should I do this at a time when people are not using the server?
I guess that I am just wondering how disruptive this will be.
also would the variables in the sysctl.conf and profile that were already there, have been loaded when the system restarted?
Just want to make sure that i dont cause any issues when these commands are run.
the last 2 commands that you have there, sysctl -p and source.
when I run sysctl -p, will those setting take effect immediately?
Also when I run source /etc/profile, I see that there are several scripts that are in there. should I do this at a time when people are not using the server?
I guess that I am just wondering how disruptive this will be.
also would the variables in the sysctl.conf and profile that were already there, have been loaded when the system restarted?
Just want to make sure that i dont cause any issues when these commands are run.
We use sysctl -p to bring the changes in effect without reboot so anytime you can use that.
Source command we use to source that particular file and read the changes without reboot so you can immediately do that it will not harm your system or otherwise when you will reboot your system next time the change will in effect after reboot and you don't need to execute both the command.
Source command we use to source that particular file and read the changes without reboot so you can immediately do that it will not harm your system or otherwise when you will reboot your system next time the change will in effect after reboot and you don't need to execute both the command.
ASKER
Thank you very much for the help. I will apply this and see if this issue happens again.
In looking through the sysctl settings I also saw that the tcp keep alive is 2 hours, so I will be lowering that as well.
Thank you for all your help.
In looking through the sysctl settings I also saw that the tcp keep alive is 2 hours, so I will be lowering that as well.
Thank you for all your help.
1. You are using openldap and fedora directory server ?
2. You said that "the mounts on the servers will become unavailable every 1-2 weeks". Mounts on ldap server or ldap server has exported some NFS share which become unavailable.
3. Mounts are local File System or Network filesystem.
4. If mounts become unavailable on LDAP server then please share output of df -h and cat /etc/fstab from LDAP server.