>>> On 12/21/2015 at 08:09 AM, "Jousma, David" <[email protected]> wrote: > All, > > We have for the longest time used a 256MB TFS for /tmp. Beauty of that is > with every IPL, the transient data is wiped out, and start fresh again. > We've very rarely ever run into filesystem space problems. However, our DB2 > sysprogs are running into problems when performing maintenance activities on > our IDAA appliances when loading new/updated code to it via their gui > interface. Seems as though that tool wants to use /tmp space between copying > data from DB2 filesystem to IDAA hardware. Some of these objects are 2G or > larger in size, so they have run into space problems, and they claim there is > no way to override what to use for temp space. > > So, I've been contemplating making /tmp a standard ZFS of sufficient size, > but am wondering how you all setup controls to manage the space?
Aside from what others have mentioned, an additional thing you could do is with every IPL, do an "rm -rf /tmp/*" command. This will replicate the previous behavior in that everything goes away with an IPL, so there shouldn't be a significant backlash from it. If you were able to get by with a 256MB /tmp before, and are only doing this for one (mis-behaving) application, I'm not sure that skulker is really going to be that necessary. Of course, time will tell. Mark Post ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
