Not a fix, but a workaround. (Well, of course every server is different - but works good for me for now....!)
Create in /etc/cron.hourly/ something like kswapd.
chmod +x kswapd
Then into it drop:
echo 1 > /proc/sys/vm/drop_caches
You'll get a lot of
[649613.095072] kswapd (29670): drop_caches: 1
in your dmesg/syslog.
But for me, this works great. Of course it depends on the bug not occurring within the first hour of the "fix" being applied (i.e. cron.hourly). And clearly this will vary from box to box. But on my AWS server (low traffic web-server with a few other odds and sods on it) this has effectively "solved" this issue enough, until we get a full fix.
Marten
Not a fix, but a workaround. (Well, of course every server is different - but works good for me for now....!)
Create in /etc/cron.hourly/ something like kswapd.
chmod +x kswapd
Then into it drop: vm/drop_ caches
echo 1 > /proc/sys/
You'll get a lot of
[649613.095072] kswapd (29670): drop_caches: 1
in your dmesg/syslog.
But for me, this works great. Of course it depends on the bug not occurring within the first hour of the "fix" being applied (i.e. cron.hourly). And clearly this will vary from box to box. But on my AWS server (low traffic web-server with a few other odds and sods on it) this has effectively "solved" this issue enough, until we get a full fix.
Sean