It's a perfectly fine and sensible thing to do. On Thu, Mar 16, 2000 at 02:32:45PM -0500, Curtis Generous wrote: > Short version: > > Q: can /var/qmail/queue live on a separate disk partition than > the rest of /var/qmail/* files? > > Long version: > > Does anyone have any reason to believe that you cannot put the > queue on it's own mount point? Either a sym link from /var/qmail/queue > --> to another disk pack or simply making /var/qmail/queue a mount > point from a different disk. Are there any dependencies that force all > of /var/qmail/* to reside on the same disk (e.g. inode numbering > issues)? My belief is that only /var/qmail/queue has issues with the > inode filenames. Correct. But not if the queue is empty at the time. There is a queue fixit script on www.qmail.org anyway. > Why: > > We attempted to do this on a two new qmail server machines (SPARC Ultra IIs > running Solaris 2.6 with QMAIL+LDAP), and after a couple of days > started to notice some performance degradation, and also that the > number of non preprocessed messages in the queue started getting very > large (thousands after a while) and the email delivery rate came to a > crawl. This was not happening on any of the otherwise identical > machines (8 of them) except that those had all of /var/qmail on the same disk. > > Unfortunately, didn't get enough time to really try to troubleshoot > this problem before we were forced into reconfiguring the two > problematic machines. The problem went away after we moved the queues > back onto the same disk and now has been working fine for several days > in the same environment (i.e. nothing else was changed). > > Can anyone think of why this problem would have occurred? I'm I can think of dozens of reasons, but their's no way of knowing whether they are relevant without any examination of the original problem in question. And indeed, moving the queue to solve a problem you don't understand is probably being optimistic. At best it probably wont get worse. > We're hesitant to try this again as it only appears after some period of > time and seems to be load related. You need to understand what's going on before you try to fix it, unless you're keen on the concept of hiring a brain surgeon to fix your stubbed toe. Regards.
