A few days ago I got myself an Antminer S3+.
Upgraded it to latest available firmware 20141126
It's basically working fine — apart from the fact that after some hours of mining at full speed it might happen the hashing speed decreases to about 10% of what it should be (i.e. ~45 GH/s instead of ~450 GH/s).
As I learnt within this forum I'm not the only one having such problems. However I haven't found a solution in this thread, thus this post.
Rebooting seems to resolve the issue, but the usual AM user has neither time nor patience to babysit his mining device all the time. And yes, it may happen he wants some hours of sleep.
In an old post regarding the Antminer S1 a Cron job is suggested:
https://bitcointalk.org/index.php?topic=588253.0The procedure:
(1) Loading up (or creating) a script ('hashcheck') to/on the Antminer
#! /usr/bin/ash
var=$(cat /proc/loadavg | grep . | cut -c 1)
if [ $var -lt 1 ]
then
/sbin/reboot
fi
(take care linebreaks are only <LF>, not <CR><LF> if you choose the
upload variant.)
(2) chmod 775 to the hashcheck script.
(3) Entering hashcheck into the list of Cron jobs:
System -> Scheduled tasks:
*/5 * * * * /usr/bin/hashcheck
The script is definitely running, it shows up every five minutes in
Status -> System log
However it does exactly nothing, even when the miner has slowed down to a crawl.
A closer inspection (looking into the Antminer's file system with WinSCP) shows that the grepped file
/proc/loadavg
always seems to have exactly 0 Bytes on the Antminer S3+, no matter in which condition the miner is running — 100% or 10%.
No output, nothing to cut, thus the reboot condition never becomes true.
Is there another way to check the miner's current condition and trigger the reboot if needed?
Or have I missed another solution for this apparently well known problem despite reclaiming / sending in the whole Antminer?
Thanks a lot in advance!
Show Posts
