Beefy Boxes and Bandwidth Generously Provided by pair Networks
Perl: the Markov chain saw
 
PerlMonks  

Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date

by sundialsvc4 (Abbot)
on Oct 12, 2015 at 21:13 UTC ( #1144589=monkdiscuss: print w/replies, xml ) Need Help??

FYI ... when I stumble-upon certain mainline sections, such as Meditations or Perl Monks Discussion, ... and for some (but not all?) threads listed on those pages ... I lately observe that the content is often many days(!) out-of-date ... until I log in.

I think that this problem varies from one major-topic to another ... e.g. Seekers vs. Meditations vs. PMD.)   Specifically, I think that I am seeing “cache artifacts” that seem to vary(!) between “headline” pages and “particular threads” within a topic ... but, not necessarily consistently.   Sometimes I think that I am seeing a “fresh” page, while sometimes I think that a “neighboring” page is, in fact, “very stale.”   (I generally suspect that “the headline page” is the one, of the two, that is “very stale.”)   Given that I dump my local browser cache many times a day, I do not believe that local-caching is the root cause of this problem.

  • Comment on Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date

Replies are listed 'Best First'.
Re: Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date
by stevieb (Abbot) on Oct 12, 2015 at 21:27 UTC

    Do you have any specific node examples or scenarios that can be tested/repro'd?

      Ah! Ha-ha-ha!!! Oh, heh. Haha, ha... oh, oh, you! ++ for the power of positive thinking and following up on help desk requests.

      $ curl -I http://perlmonks.org/?node=Recently+Active+Threads 2>/dev/nu +ll | egrep -i 'date|cache|modified' Date: Mon, 12 Oct 2015 21:43:54 GMT $ curl -I http://perlmonks.org/?node=Recently+Active+Threads 2>/dev/nu +ll | egrep -i 'date|cache|modified' Date: Mon, 12 Oct 2015 21:43:57 GMT $ curl -I http://perlmonks.org/?node=Newest+Nodes 2>/dev/null | egrep +-i 'date|cache|modified' Date: Mon, 12 Oct 2015 21:44:07 GMT $ curl -I http://perlmonks.org/?node=Newest+Nodes 2>/dev/null | egrep +-i 'date|cache|modified' Date: Mon, 12 Oct 2015 21:44:09 GMT
        At least I didn't ask "did you turn it off and back on again?" :)
Re: Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date
by jdporter (Canon) on Oct 15, 2015 at 18:49 UTC
    I lately observe that the content is often many days(!) out-of-date ... until I log in...

    I don't know if the problem is on your end, or somewhere in between, but it's not on our end, because PM doesn't cache any output. (There are a few pages which are "cached", such as Selected Best Nodes, but you don't seem to be talking about any of those.)

    If you would like to take some screen shots the next time you see it happen, and send them to me, I'd be happy to take a look.

Re: Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date
by Anonymous Monk on Oct 12, 2015 at 23:35 UTC
Re: Cached(?) pages when not-logged-in sometimes appear to be very seriously out-of-date
by sundialsvc4 (Abbot) on Oct 27, 2015 at 15:58 UTC

    I observed this behavior again yesterday evening (Eastern Time US), and was able to experiment with it a little bit:

    • The list of comments on three threads (all the threads that I checked) was incomplete, in that a complete list of comments was not presented.
    • The appearance of the page was entirely correct, although its display of the thread content was stale.
    • Emptying the browser cache did not change the result.
    • Examining the page on another browser also did not change the result:   it was stale there, too, and identically so.
    • Upon logging-in, the content of the threads was now current.   Additional comments appeared which had not been there before.
    • The other browser, which was not logged-in, continued(!) to display the stale content:   the pages as displayed by the two browsers, one logged-in and the other not, was now different for these threads.   (I only “refreshed” the page:   I did not re-request the URL, nor did I think to flush the cache on the second browser.   I do know that this browser is set to flush its cache on program startup and shutdown.)

    Although this behavior is only an annoyance (along with the occasional “disappearing session,” which I somewhat suspect might be related), I offer this finding in hope that the gods might be able to find and fix it.   (The last bullet-point makes me wonder if it really is or isn’t a problem with a down-stream caching server, and I unfortunately did not do an exhaustive test with the second browser when the opportunity to do so had presented itself.)

      Thanks. It sounds like you've got a protocol which should make it possible to reproduce your observations. Next time you observe the behavior (or this time, if you still have the browser windows open from before), could you please note, and report here, two items of information: (1) the thread (url) where you see the problem happening; and (2) more importantly, the specific comments (by node_id) that appear in the logged-in view of the thread but not in the logged-out view. Just a handful should suffice, if the number of comments in that set is large. Thanks!

      I reckon we are the only monastery ever to have a dungeon stuffed with 16,000 zombies.
      foo

        I suspect this is to do with the 'Replies headers depth' & 'Replies text depth' settings.


        With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
        Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
        "Science is about questioning the status quo. Questioning authority". I knew I was on the right track :)
        In the absence of evidence, opinion is indistinguishable from prejudice.

        I do not even know what these settings are:   in all these years I have never used anything but the default settings, whatever they are.

        The curious thing about this problem is that it definitely is sporadic.   I got seriously-slammed for reporting it(!) on October 12th, and did not see it again for 16 days.   Yesterday was also (coincidence?) the most recent date that I had a “disappearing session,” although I have learned to watch for this and caught it in time.   What really startled me is when the second-browser, after Refresh, did not show the updated page.

        I do not know the intricacies of how Anonymous Monk is implemented, but I presume that the system does see it as “a real user,” and I idly wonder if there could be something amiss with that (obviously, heavily-used, and shared) user-table entry.   But my well-worn debugger’s hat tells me that, when software screws-up, it ought to be reproducible at-will if you just know what to do.   I am also very surprised that no one else chimed in to say that they had seen this, too.   (Although I understand that they might decline to do so, in order to avoid snarky comments and down-votes.)   So far, I have not had an epiphany as to what the problem might be.

      If this report came from a non-technical user I would be surprised at how good and thorough it was. Since it came from an "experienced" dev who has gotten quite a few pointers about diagnosing things like this more deeply, from me and others here, it is lazy and unhelpful. No timestamps? No browser Ids? No request/response headers or full captures? No internet provider/network info? Nothing new. Nothing for an interested party to dig into.

    A reply falls below the community's threshold of quality. You may see it by logging in.
    A reply falls below the community's threshold of quality. You may see it by logging in.

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: monkdiscuss [id://1144589]
Approved by Corion
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others taking refuge in the Monastery: (3)
As of 2019-04-20 18:24 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?
    I am most likely to install a new module from CPAN if:
















    Results (110 votes). Check out past polls.

    Notices?
    • (Sep 10, 2018 at 22:53 UTC) Welcome new users!