Using Amahi as DHCP/DNS may have caused client instability?

[vitocorleone]

As of the time of this writing, it's possible that it was a coincidence that, after installing Amahi and turning off DHCP in the DGL-4300 router, both client computers (XP Pro) became "unstable"; the PC (new) would freeze up when opening/using Firefox (system was 99% stable before Amahi) such that it had to be restarted and the laptop had Firefox crash/freeze up a few times as well, though I don't think I had to restart it.

After having Amahi running for 2 days or so I turned DHCP back on in the router and, so far (going on one day), zero issues. I'll update this thread if the PC and/or laptop have any issues. Again, I can't attribute the instability to using Amahi as of yet, but wanted to post my experience to see if anyone else has encountered something similar.

I still have Amahi running, but just as a file server for now.

Server: P4 3.06HT w/1GB ram on Asus P4Bfive33 (my five key on my laptop just stuck!) and Intel 1000MT Pro NIC
PC: Q9450 w/4 GB ram and built-in Realtek gigabit NIC(s) on Gigabyte GA-EX38-DS4
Laptop: Dell D600 w/1 GB ram and built-in Dell NIC using broadcom to connect wirelessly

Main router: D-Link DGL-4300 (wireless not enabled)
Secondary router: Linksys WRTfifty4GL (wireless only) - hard-wired to DGL-4300

Also on the DGL-4300 is my Lexmark laser printer

Thoughts? Ideas?

[cpg]

one can never say it's not possible. we have hundreds of users using their servers 24x7 for many many months now. stability overall is proven.

however, there could be issues like intermittent failures in the dhcp or the dns server for some reason, though this is very uncommon. we do have provisions to restart them if they fail, however, they could consistently fail and be restarted, for some reason.

let's assume that is the case.

in that case, one suggestion is to look in the logs of the amahi HDA (at the end of /var/log/messages). this will tell you if the servers we being restarted by monit (our server process monitoring system).

if you do not seem anything unusual there, then perhaps changing your DNS to your ISP would be another test you could conduct, if it's reproducible. however, we do not have easy (read: GUI) support for this at the moment.

so i would recommend going to the IRC channel and seeking some handholding there, as it's not entirely as simple as it should be.

[vitocorleone]

Thanks - I'll check out the log file(s).

Going on 2 days now and not a single hiccup since returning DHCP duties to the router. Again, though, for all I know it could've been an issue at the ISP (Comcast) at the time... I'll try to determine as I am curious (but time limited).

EDIT: Getting a lot of these in the log

Dec 31 23:11:56 localhost monit[2370]: 'mt-daapd' process is not running
Dec 31 23:11:56 localhost monit[2370]: 'mt-daapd' trying to restart
Dec 31 23:11:56 localhost monit[2370]: 'mt-daapd' start: /etc/init.d/mt-daapd
Dec 31 23:12:26 localhost monit[2370]: 'mt-daapd' failed to start

and (i replaced ip addresses with xxx's)

Dec 31 23:13:06 localhost named[859]: too many timeouts resolving 'xxx.xxx.xxx.xxx.in-addr.arpa/PTR' (in '.'?): disabling EDNS

and

Dec 31 23:11:52 localhost named[859]: unexpected RCODE (SERVFAIL) resolving 'xxx.xxx.xxx.xxxin-addr.arpa/PTR/IN': xxx.xxx.xxx.xxx#53

these kind of events were logged for a couple of days, repeatedly and constantly

[cpg]

hi vito

these messages are annoying, but thankfully harmless.

the mt-daapd one is a discrepancy in writing the pid file. it cannot get restarted because it's already running, it's just that the pid file did not get written properly due to a race condition somewhere.

disabling EDNS is when your provider (like mine, comcast also), does not provide extended dns (EDNS) info.

the other one happens for a variety of reasons. many sites are not resolvable. this can happen if you receive email or have spam processing, which checks validity of headers, or if you have skype, which tries to resolve the name of peers ... i see them all the time.

we should look into filtering these two last messages all the time (and fixing the first one!).

it looks like your dns and dhcp server in the amahi hda is ok. remember after you switch dhcp server, your clients' network interfaces need to be "repaired", i.e. their lease be released a a new one be requested, with the amahi hda respond to it (or you can also reboot your clients).

carlos

[vitocorleone]

Hmm. Well, I'm going to give my current setup a couple more days to see if any issues come up and then switch back to using Amahi (I do restart the clients so they'll pick up their new IP addresses) for awhile. If I run into more of the same as I had before I'm willing to PM or email you the logfile at that point so you can dig through it since you can probably spot "real" issues better than I!

FYI - with 2 XP Pro clients it's been working great as a file server. I just need to add more capacity (ie remove a 320gb drive from my current computer that already has 2 x 620 GB drives and put it in the server box).