Login slowness from LDOM to ZONE

Need assitance in troubleshooting steps . Below is the issue

I have LDOM which has 4 zones on it . One zone logging into the server takes 13sec where as other zones are quick .

Troubleshooting steps which I did

  • Checked /var/tmp . Not many files
  • We have LDAP but thought that wouldnt be an issue as I am logging from LDOM to Zone as zlogin <hostname>
  • If I login into zone directly the user logging is very slow so suspected to be a DNS issue and verified resolve.conf and verified the DNS IP's
    -nsswitch.conf
root@hrappt53:~# cat /etc/nsswitch.conf |grep dns
hosts:  files dns
ipnodes:        files dns
  • Restarted zones to see if the issue resolved no use

I am running out of options . Need help

Which version of Solaris?

When does the 13 secs get wasted? Doesn't give you a quick login prompt? Doesn't react quickly after the password is input?

When you experience the delay, how are you trying to login exactly? If you login to the global zone first does it then login at full speed when to connect to zone from there with 'zlogin'?

Which version of Solaris?
Solaris 11.3

When does the 13 secs get wasted? Doesn't give you a quick login prompt? Doesn't react quickly after the password is input?
I get login prompt but after I enter password it doesnt react quickly . Thats where it take couple of sec

When you experience the delay, how are you trying to login exactly? If you login to the global zone first does it then login at full speed when to connect to zone from there with 'zlogin'?
Logging to Global zone is quick . But when logging into zone and that too only for this zone and rest 3 zones are fine. I suspect something going on with this zone .

This could be caused by a number of things. What follows is not an exhaustive list but let's highlight a few.

  1. Is that zone really busy for some reason? Try running performance stats (sar)
  2. Is that zone being strangled by some resource constraint imposed when the zone was configured. Compare config with other zones. (zonecfg)
  3. Are these zones whole root or sparse? Is usage of the global zone affecting this zones performance?
  4. How is the user authentication carried out? Locally or remotely? Compare with other zones.
  5. Can the login delay be interrupted? Can you break out of the delay time by hitting Ctrl-C? Does the prompt appear immediately if you do that? Or is the process uninterruptable?
  6. The zone may be doing a reverse lookup to a node that isn't responding. Compare resolv.conf files between zones. Is the server(s) reachable? Working with one login client can you fix the problem by putting it's name & ip into that zones /etc/hosts making it instantly resolvable.
  7. Does that zone have exclusive use of its own network interface? If it's sharing an interface (with the global zone for example) is traffic generated by the other zone delaying a reverse lookup?

Let me stop there for the moment. It's difficult to provide a definitive answer without you doing some of your own research. Do please post back your progress.

We're all here to help.

Finally it worked :b:

Shutdown all the application reboot the zone and tested the login again and it worked

Is it always 13 sec or can be longer (even forever) ?
In most occasions such errors were no memory to fork or resolving (DNS) issue.

If it's always 13 sec, should be DNS releated.
Rebooting the zone probably applied the correct configuration in /etc/resolv.conf or one of the entries in resolv.conf started working again :slight_smile:

Regards
Peasant.

The issue is back again after few days but found another solution

in /etc/profile there is a quota which checks the NFS mounts as well. My assumption that it is checking the quota of the mounts which is taking the user very slow login

at the user prompt try

touch .hushlogin

this will surpass checking the quota

Alternatively you can mount NFS with the noquota option.