Yes, I know.

In my observations, it is mostly the 216.* machine that is slow resp. has wonky/faulty/weird network routing problems, so my plan is to remove it from our DNS rotation. If anybody is aware of reasons for not doing that, please speak up.

Hello Corion and thanks,

> it is mostly the 216.*

not always:

perlmonks.org @ https://66.39.54.27/index.pl?           200 OK 26.5322
+ Kb (27169 bytes) in 12.627326 seconds 2.1012 Kb/s
perlmonks.org @ https://216.92.34.251/index.pl?         200 OK 26.5322
+ Kb (27169 bytes) in 9.580532 seconds 2.7694 Kb/s
[download]

We used to have 3 machines iirc and one was removed. Having at least two machines is not the minimum redundacy needed? ..well the load nowadays is not that big and the DNS roundrobin is not the best at all as HA solution.. simply is not.

What Tux noted below is really weird: it means it was reachable only from some country/network? UPDATE ..and yes choroba confirmed the same behaviour noticed previously

L*

There are no rules, there are no thumbs..
Reinvent the wheel, then learn The Wheel; may be one day you reinvent one of THE WHEELS.

What Tux noted below is really weird: it means it was reachable only from some country/network? UPDATE ..and yes choroba confirmed the same behaviour noticed previously

That sounds more like a routing or firewall issue and less like a server issue.

Alexander

--
Today I will gladly share my knowledge and experience, for there are no sweeter words than "I told you so". ;-)

It is not good for our community.

What also bothers me is that quite often in the past week, *some* IP's seem to be unable to reach PM at all! choroba and myself where on IRC sharing the moments when PM was/was not reachable form home and/or $work network, and under the same ID, both from our home networks were not able to use PM at all.

It is not good for our community.

I wholeheartedly agree. However, it is also not a good look for Pair. I assume that these problems have been brought to Pair's attention but so far they have not been able to solve them. Perhaps we should be sounding out alternative hosting providers?

---

🦛

this morning something very weird happened with pm.

In the Tk chat client by choroba I read as last message [choroba]: Also, I was surprised how cheap things are outside the tourist areas.... and I posted [Discipulus]: good morning and nice camelday! and the message appeared in the Tk client.

But the site was unreachable:

Wed Feb 12 09:35:51 2025 checking perlmonks.org at 66.39.54.27 216.92.
+34.251
--------------------------------------------------------------------
perlmonks.org @ https://66.39.54.27/index.pl?   500 Can't connect to 6
+6.39.54.27:443 took 15.265145 seconds (timeout 15 seconds)
perlmonks.org @ https://216.92.34.251/index.pl? 500 Can't connect to 2
+16.92.34.251:443 took 15.135559 seconds (timeout 15 seconds)
[download]

At 9:42 GMT+1 both server still 500 and I posted [Discipulus]: still.. really weird and it appeared in the client.

At 10:50 GMT+1 both server were 200 OK, I posted succesfully in the chat: [Discipulus]: now both 200 OK

last hour of cb says: Last updated 2025-02-12 09:48:49 UTC - chatterbot    *** The monastery has been quiet for 00:58:10 ***

mini-cb60.datenzoo.de is empty but http://pthbb.org/cb/last.cgi shows:

2025-02-12 04:51:00  Discipulus
now both 200 OK

2025-02-12 03:47:53  Discipulus
still.. really weird

2025-02-11 07:06:46  choroba
Also, I was surprised how cheap things are outside the tourist areas..
+..
[download]

..so missing two messages I succesfully sent from the Tk client and appeared in it.

I hope this helps to debug the problem

L*

chatterbot was pretty much unable to communicate with PM. It seems it was occasionally able to update some pages, but it was pretty much missing out on getting most of the posted chatterbox messages:

# select logtime, username, message from chatterbot.inbox where logtim
+e > '2025-02-11 14:30' order by logtime;
          logtime           |    username    |                        
+                message                                        
----------------------------+----------------+------------------------
+---------------------------------------------------------------
 2025-02-11 14:32:44.782183 | choroba        | Getting timeouts...
 2025-02-12 07:59:24.645271 | Discipulus     | good morning and nice c
+amelday!
 2025-02-12 08:50:38.607121 | Discipulus     | still.. really weird
 2025-02-12 09:51:14.57602  | Discipulus     | now both 200 OK
 2025-02-12 16:14:51.780973 | hippo          | :-( This is painfully s
+low. Will try again tomorrow.
 2025-02-13 15:41:13.509351 | marto          | hey all
 2025-02-13 16:31:43.35478  | hippo          | ... aaaaaaaaand we're b
+ack. Have the bots given up or are they just having their tea?
[download]

At least on and around the 12th of February, this doesn't look like a technical problem with PM per se. The failure rates seem to rise over some hours, see pm server stats. If PM is unreachable, you can also view the stats on my external site.

Please note, though, that chatterbot only logs that something has gone wrong during a web request. It currently doesn't log what exactly has gone wrong. Also, chatterbot runs as a cyclic executive¹. It tries to do the webcalls at a certain frequency, if some (or all) webcalls take a long time to complete or run into timeout, the number of webcalls per hour sharply decreases.

I technically do log the status and time taken for every webcall and use that to calculate the per-hour stats for the web pages. I doubt this will be much help, but i certainly can provide that data if it seems useful.

---

¹ A form of cooperative multitasking within a single program. Wikipedia. My server runs quite a few of these "workers". All of the chatterbot code is in a single worker, because it doesn't make sense to parallelize PM tasks.

PerlMonks XP is useless? Not anymore: XPD - Do more with your PerlMonks XP
Also check out my sisters artwork and my weekly webcomics