Installing cron in boot2docker-xhyve to fix clock drift
Closed this issue · 15 comments
From what I can see, there is no cron installed in boot2docker-xhyve, is this correct?
One interesting bug is that the clock drifts while the host machine is asleep; this can cause things like AWS APIs to fail within Docker.
An easy fix is to run sudo ntpclient -s -h pool.ntp.org in the boot2docker VM. Ideally I'd like to put this in a crontab, however without cron in the VM, I am rather at a loss!
Any ideas?
boot2docker/boot2docker has /usr/sbin/crond and /usr/bin/crontab.
docker@boot2docker~$ cat /var/spool/cron/crontabs/root
# restart ntpd to combat laptop sleep + VM pause
0 * * * * killall ntpd > /dev/null 2>&1; /etc/rc.d/ntpd
Or just $ sudo reboot would be fine.
Ah beautiful, I couldn't find the crontab for the life of me. Thanks!
Sent from my iPhone
On 2 Oct 2015, at 11:26 AM, A.I. notifications@github.com wrote:
boot2docker/boot2docker has /usr/sbin/crond and /usr/bin/crontab.
—
Reply to this email directly or view it on GitHub.
xhyve should take care of the clock drift during host sleep eventually.
@ailispaw It seems that it is incorrect even using NTP?
That's after a fresh boot of the VM. Any ideas?
@ailispaw Brisbane/Australia (UTC +10).
The main issue that this causes is breaking the AWS S3 API, as it whinges about clock drift being too large :(
@girvo
What about the time after executing sudo /usr/local/bin/ntpclient -s -h pool.ntp.org?
Will it adjust the clock correctly?
Oh you have already mention it at the first comment. Sorry.
I think that you can replace ntpd with ntpclient in the crontab and set more frequently.
Or you can run the following command in the host after resume. I think this is the easiest way.
$ make ssh -- sudo /usr/local/bin/ntpclient -s -h pool.ntp.org
That's after a fresh boot of the VM. Any ideas?
I have no idea... I've never seen that before.
Is this related to boot2docker/boot2docker#1082?