Problem on dcc 1.3.30 - Continue Not Asking DCC...

Vernon Schryver vjs@calcite.rhyolite.com
Fri Mar 10 14:46:04 UTC 2006


> From: Breno Moiana 

> At this very moment, the server is unresponsive. I will try to provide diagnostics info, and then I will do the things you suggested (update, remove the anon, etc):
>
> 1. The DCC server
>
> 1.1
> # cdcc "host localhost; stats"
> localhost 127.0.0.1,512
>         server-ID 1004  09:42:39

That says the server is working and reasonably happy.

> 1.2
> # cdcc info

> 127.0.0.1,-                 RTT-1000 ms  anon
> # *127.0.0.1,-                                               OiComBR ID 1004
> #     100% of 32 requests ok   50.18-1000 ms RTT        50 ms queue wait

That also says your server is resonding.
There should have been something like this about skipping asking
at the start of the `cdcc info` output:

    # Re-resolve names after 09:25:44  Check RTTs after 07:40:48
    # 1 total, 1 working servers
    # skipping asking DCC server 54 seconds more


> 2.3 ps:
> This might be interesting. /etc/init.d/dcc start loads three processes: dccd, and two dccifd:

The extra dccifd process does nothing except wait for the active
dccifd process to quit.  If the active proces quits with core-dump
or other major problem, the extra process starts a new active dccifd
process.  of course, with rate-limiting.


>     0 bad IDs  407 passwds  37 error responses  5929 retransmitted

Those values are unchanged from the previous `cdcc info`, but they are
odd.  407 times since "Feb 21 10:53:16.629997 BRT" a DCC client has
tried to talk to your server with the wrong client-ID and password
combination.  ALso, there have been 5929 requests that the server
recognized as retransmitted.  That is strange when the client is on the
same computer.  Then there were the 37 otherwise bad requests.

It would be interesting to see if any of those values increase
when things go bad.

Note that `cdcc "stats clear"` resets all of the counters.


> >The second server would be need only when the first server is not answering.
> >
> Oh, okay. I thought you meant that with 3 million messages, I would be 
> overloading something that should be split between two servers for 
> balancing purposes, not only for redundancy and service availability.

Several of the public DCC servers handle more than 3 million requests/day.
One has done 11.2 million in the last 24 hours.


> Yes, I had the impression this was happenning. My point wasn't the 
> countdown per se, but the fact that it both stops and gets back to 
> responding without any interference.

I have a theory that might explain excessively high counts.  It may be 
possible for several mail messages to almost simultaneously suffer
problems talking to the server and each double the counter.
I'll have to try to figure out a way to test that theory.


Vernon Schryver    vjs@rhyolite.com



More information about the DCC mailing list

Contact vjs@rhyolite.com by mail or use the form.