Strange behavior slows server down

21 views
Skip to first unread message

Dave S

unread,
Mar 30, 2022, 5:53:15 AMMar 30
to web2py-users
Recently, I've been seeing one of my test web2py servers get bogged down, and sometimes fail to respond to the user.  When this happens, I find the web2py.log file is getting filled with lines like

2022-03-29 17:38:31,480 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:38:46,710 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:38:47,776 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:39:00,003 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:39:21,467 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:40:21,421 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:41:21,367 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)
2022-03-29 17:42:21,348 - web2py.scheduler.Dave-PC#8844 - DEBUG - defining tables (migrate=True)

This continues even if the user (me) stops making requests.  This being Windows, I launch the server using the little GUI tool that includes a traffic monitor (seismograph of incoming requests), and that shows a regular tick that isn't connected with my clicking any links.

The server version info:

web2py™
Version 2.21.1-stable+timestamp.2020.11.28.04.10.44
Python
Python 2.7.16 (bundled)
The front end is the bundled Rocket code.

The client is Chrome:
Version 99.0.4844.82 (Official Build) (64-bit)

Both client and server are on a laptop:
Edition        Windows 10 Pro
Version        21H1
Installed on        ‎11/‎15/‎2020
OS build        19043.1586
Experience        Windows Feature Experience Pack 120.2212.4170.0

I do have a model for a scheduler task, but I usually don't fire up the scheduler process.  If I change the filename for that model to scheduler.py.hidden, I don't get those scheduler log entries, but the seismograph still shows an event about once a minute.

httpserver.log says it's the same three requests each time:

92.168.4.21, 2022-03-30 02:42:21, GET, /admin/todolist.load, HTTP/1.1, 303, 0.043000
192.168.4.21, 2022-03-30 02:42:21, GET, /admin/index.load, HTTP/1.1, 404, 0.081000
192.168.4.21, 2022-03-30 02:42:21, GET, /badlist/appadmin/hooks, HTTP/1.1, 200, 0.284000

With older versions of web2py and Chrome, I've seen what I call "cache thrash"; where Chrome seemed to be issuing a blizzard of requests to refresh the cache, and web2py couldn't respond to anything else.  This would continue until I stopped the server, which often required a process kill.

Am I see a less vicious form of cache thrash?

I became aware of this current version of the problem when using a "public" port that is visible on the local network (having bought a new laptop), but I don't think that's the source of the problem, and the earlier version of the problem was only using 127.0.0.1 (localhost).

/dps


Dave S

unread,
Apr 1, 2022, 6:17:14 PMApr 1
to web2py-users
So far, I'm not seeing the problem with Opera as the local client, or Chome across the network, but I haven't stressed things yet.

BTW, have others of you seen a problem [unrelated to web2py] where Chrome on W11 Home doesn't properly redraw the page after coming out of sleep?  OS Build 22000.556 and Chrome  99.0.4844.84 (64-bit).

/dps

Dave S

unread,
Apr 6, 2022, 3:23:42 AMApr 6
to web2py-users
Another off-topic complaint:  The current Google Groups version (for about a year now) will show the correct number of posts in a thread when you're looking at the conversations page, but when you open the thread, it only shows posts you've already read (in a previous read) and has a banner on the bottom for "2 new posts ..... read?"

This can hours later.

/dps "yeah, I'm griping"
Reply all
Reply to author
Forward
0 new messages