With Red Hat 5.9 (Starbuck) All crons are ran at boot. That causes some problems on my system. I add the crons in rc.local so when it reboots it automatic. But, then all the crons run after crond comes up, cause crons to spawn bad processes. 5.2 didn't do this.
What sort of 'bad processes' get spawned?
For instance, I run a Linux Quake 2 Server. So, I have rc.local start the quake server. Then I have multipile crons to reboot the quake server, and check to see if it is running etc. But, on reboot, the quake server comes up in rc.local. Then soon as crond comes up all my crons are ran. Thus cause my cron to reboot quake once a day starts, kills the server, then the another script sees the server isn't running so starts a quake server. So, by the end of reboot, I have like 6 quake servers running. And with the same configs, and crons I had in 5.2. 5.2 the crons where flawless.
The new cron scheme is the right way to do things. If you have a laptop for instance, and it is off for some time, when it came back on before, there would be cron jobs that had been "skipped." This is a bad thing. Cron jobs that are scheduled to have been run should run when the machine comes back up. We understand that this is a policy change on our part, but internal discussion as well as discussion with the LSB group has lead to the change. We suggest that you rewrite your local cron job.The new cron scheme is the right way to do things. If you have a laptop for instance, and it is off for some time, when it came back on before, there would be cron jobs that had been "skipped." This is a bad thing. Cron jobs that are scheduled to have been run should run when the machine comes back up. We understand that this is a policy change on our part, but internal discussion as well as discussion with the LSB group has lead to the change. We suggest that you rewrite your local cron job.
I believe crons should run when scheduled to. Not all on boot. So, if I schedule a cron to run once a day at 5 in the morning, I don't want it to run when I reboot the machine at 2pm.
I believe that I am wrong -- the cron behaviour has NOT been changed to the new behaviour (the discussion is still pending). The cron we ship is the same that was included in 5.2, it has not changed. I'm not sure why you had the problem, but cron has not changed. Perhaps your time and date became skewed.
Ok I see. So, 5.9 is still the same set up as the old 5.2 cron setup. See, I wrote this a bit too early, see, here's what I have in rc.local, after the normal stuff this is gereneral too. quake2 deathmatch quake2 ctf crontab /usr/util/cron ipchains [my firewall stuff] getdate Now, here's the problem. The bios date is correct, but linux reads the date a day early. So, like a co-admin where the box is located reboots it when he can't figure out what's wrong (nice huh?) and the logging routines for the deathmatch get messed up. So, getdate, sets the machine to the correct date from a different machine. That were the cron job get messed up. But, to make a long story short, 5.9 looks awesome can't wait till 6.0. We had to revert back to 5.2 at the moment. When I tried installing rpm's on 5.9 I got dependancies that were wrong. And the quake servers run fine until they get a lot of traffic then die. Don't know why. nothing has changed. I was thinking the new lib's or dep's.
Commit pushed to master at https://github.com/openshift/openshift-docs https://github.com/openshift/openshift-docs/commit/a850e36e0483bf61f5b6accfb194ed08e1f7d03c Merge pull request #1662 from bfallonf/issue_1656 Added note box about template location as per issue 1656