Topic: ANUBIS - a CGMINER Web Frontend - page 10. (Read 83112 times)

Hey P_Shep, you left a bunch of bare carriage returns in your edited files.  Did you edit them in a DOS editor and then upload them without conversion?  It's playing hell with DIFF and editing locally on a unix machine.

Vast amount of changes in Anubis, go have a play

https://github.com/pshep/ANUBIS

I'm going to be putting some work into Anubis this week related to historical tracking. I suck at charting though, so once the data is collected, maybe someone can write some code to make good charts?

Thanks for the work Inaba and the other suggestions here. I'm on vacation for the next week, so will get back on it when I'm back.

Regarding the multi-rig changes... I did think about that, but you need to guarantee that the pools are ordered the same on all rigs, something that is outside anubis' control, things could get out of sync. Doable though. I'll have a think.

I submitted the pull requests... pretty handy I can do it online.

.htaccess alone won't solve the issue, as P_Shep's anubis allows finer control over the miners.

You can actually edit a file and make simple pull requests right on github.com.  No need to manually fork and all that.

Also line 32 of index.php should probably be changed to:

$result = $dbh->query("SELECT * FROM hosts ORDER BY name ASC");

I may setup a git and issue pull requests, but it's not worth it for the small changes I've needed to make so far.

A great project and one I have enjoyed making some minor changes to.

The largest annoyance is the rig communication seems to be blocking.  If I have a rig which is down Anubis takes 10 sec trying to establish a connection meanwhile the entire webpage is unresponsive.

The default DB schema for ANUBIS has an error.  You have the "port" field set to smallint (6) and it's signed - it should be set to a bare minimum of MEDIUMINT (6).  Otherwise you are limited to ports < 32768

Either that or make smallint unsigned.

What I think would be really great is to have the miners organized into groups and be able to change them all in one shot. Change a pool -> click. They all go. Go from failover to roundrobin or whatever-> click done. Clicking through each one is not very fun.

this is a setting in .htaccess or file permissions on the webserver. It has nothing to do with anubis itself.

That's an easy one, I'll fix that with my next commit.

One potential bug ....

It looks like discards % is (discards / accepted) but this is not a valid metric.
Discards are discards getworks.  so discard % should be (discards / getworks).

For users where accepted ~= getworks then error is small but for pools like p2pool where due to LP a large number of getworks are discarded and shares are > 1 difficulty (thus lower accepted) it results in some "weird" % like 287% discarded.

I'll think about that, but I think it'll be more trouble then it's worth.
Maybe I'll get bored one day and do it.

Just an idea, I think it would be nice if there was some way to have the site require a login or something for any webpage commands to be executable.

Sometimes I just want to show friends of mine my ANUBIS page without giving them the potential to blow up my mining rigs.

If you are restarting cgminer, the issue is most likely the side effect of the socket being in the TIME_WAIT state as is somewhat considered mandatory by internet protocol standards.
There is a way around it in Linux by doing something considered bad:
messing with net.ipv4.tcp_fin_timeout

There's also the socket option SO_REUSEADD that I STILL haven't got around to testing if there are any issues with using it in cgminer.

+1

That's a good one. I'll add it when I next do an update. I'll make the timeout configurable in the config file.

I just started trying out Anubis; thanks to ocminer and p_shep, its very nice.

One odd thing I noticed... if one (or more) of my rigs is hung or in the process of going down, it takes a really long time (more than a minute) to load the Anubis home/hosts page.  If a rig is down, the page comes up quickly (less than one second) showing 'Connection refused' or 'Operation now in progress'   I have 20 rigs on the local network.  Anubis is served from a non-mining machine.

I tried lowering the timeout in the socket_set_option on line 81 of func.inc.php and that seemed to help a little, but I think whats happening is that cgminer is hanging on exit (when shutting down due to a sick or dead card) and not responding to Anubis.  Anubis then sits waiting for a response from that rig which may never come.

I added the following receive timeout to line 82 and it seems to have fixed my problem:

socket_set_option($socket, SOL_SOCKET, SO_SNDTIMEO, array('sec' => '1', 'usec' => '0'));
socket_set_option($socket, SOL_SOCKET, SO_RCVTIMEO, array('sec' => '1', 'usec' => '0'));

I'm not sure what your patch policy is, and I'm not familiar enough yet with git/github, so forgive me if this is the wrong place to submit this.

I suppose the timeout values could be variables in the config if you wanted, and 1s is probably not long enough for people monitoring rigs across the internet, but this seems to be working for me and I hope it helps someone.