Summary: Most heavy users on our system have been experiencing some slowdown of e-mail for the last couple of weeks. This was up until recently intermittent however it is now a problem that occurs during peak usage times. i.e. between 11am and 5pm or so.
The cause for this issue appears to be because of a number of factors, some of which are out of our control.
Within our control our storage array that is well speced doesn't appear to be doing the business. i.e. under heavy slow it's quite slow to respond. We've taken this up with the vendor in question, however we're building a second SAN to replace it. We hope to put the new storage in place during the weekend of October 17 and 18th. This should give much better performance over all and will also stop the mail delays.
Outside of our control we have ldap and whosond which are services parallels use to maintain certain databases of user information that qmail and couier-imap use to get information about users. Each time these have issues it causes problems with mail delivery, this compounded by the storage array being slow is the cause of the problems.
We've also got ram upgrades which we're going to put in place tonight on all the mail nodes,
because we're clustering them there'll be minimal impact during the ram upgrades.
Please bare with us while we get this issue resolved.
The cause for this issue appears to be because of a number of factors, some of which are out of our control.
Within our control our storage array that is well speced doesn't appear to be doing the business. i.e. under heavy slow it's quite slow to respond. We've taken this up with the vendor in question, however we're building a second SAN to replace it. We hope to put the new storage in place during the weekend of October 17 and 18th. This should give much better performance over all and will also stop the mail delays.
Outside of our control we have ldap and whosond which are services parallels use to maintain certain databases of user information that qmail and couier-imap use to get information about users. Each time these have issues it causes problems with mail delivery, this compounded by the storage array being slow is the cause of the problems.
We've also got ram upgrades which we're going to put in place tonight on all the mail nodes,
because we're clustering them there'll be minimal impact during the ram upgrades.
Please bare with us while we get this issue resolved.