Friday, April 19, 2013

RAM Disk Full for ESXi on HP servers

Issue: when trying to do a vMotion they get an error about the ramdisk being full.
what we tried:
http://communities.vmware.com/message/2033719
http://communities.vmware.com/message/2026032
kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2001550
checked for the wtmp but did not find it.
What worked :
Download and install "HP Agentless Management Service Offline Bundle for VMware ESXi 5.x"
http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15351&prodSeriesId=5177957&swItem=MTX-0121a632c89f4e75b3d7c42103&prodNameId=5177958&swEnvOID=4141&swLang=8&taskId=135&mode=5
Why :
To avoid filling the RAM disk, AMS is now creating its log file “hpHelper.log” in the configured scratch location (default of /scratch/log directory).  In addition, a soft link to the log file is created in the path /var/log.  The hpHelper.log file may grow to a maximum size of 10MB and a file rotation algorithm has been implemented with support of up to 10 archived hpHelper.log files.
courtesy:colleagues@hp

Monday, April 15, 2013

Enabling ssh for non root users in esxi 5.x

Issue: User named 'test' needs ssh access to the esxi 5.x host.
Resolution: edit /etc/security/access.conf
login as root via ssh to the host and
vi /etc/security/access.conf

+:dcui:ALL
+:root:ALL
+:vpxuser:ALL
-:test:ALL
+:UserToGiveSSHTo:ALL
-:ALL:ALL
 and change it to

+:dcui:ALL
+:root:ALL
+:vpxuser:ALL
+:test:ALL
+:UserToGiveSSHTo:ALL
-:ALL:ALL
now you can log in to the esxi 5.x host using the username test but this is temporary since the changes to the file /etc/security/access.conf will be reverted to the original settings after a restart of the management services or the host reboot. You can take care of that by backing up the settings to the startup configuration by a simple command
/sbin/auto-backup.sh

Thursday, April 4, 2013

Ideal iSCSi potgroup nic teaming policy for ESXi 5.x

As we all know that in esxi 5.x we have a problem (from ages of course) that if we keep both the Nics as active active in the iscsi portgroup then it will drop packets and you might start having some problem with the storage connectivity.
The ideal workaround that i have come across myself is
1st method.
create 2 portgroups in namely iscsi1 & iscsi2 in vswitch1.
In the portgroup iscsi1 keep the vmnic1 as active, vmnic2 as unused.
In the portgroup iscsi2 keep the vmnic2 as active and vmnic1 as unsed.
optional:You can have this both the iscsi portgroups on the same or different vswitch.
2nd method.
2 iscsi portgroups with only 1vmnic per iscsi portgroup and both of them kept active.