20

The search functionality is broken, it seems, e.g. https://meta.stackexchange.com/search?q=test only loads after more than a minute of waiting. This happens on multiple sites, e.g. also on Stack Overflow.

  • 1
    Confirmed here, using FF on Mac, I get the offline for maintenance page after a while – Luuklag Nov 16 at 11:22
  • 1
    The search of the Help center is also impacted, as is chat search. – rene Nov 16 at 11:25
  • 1
    search completely failed on meta mathematics whereas mathematics search queries dont return results – Queenie Goldstein Nov 16 at 11:39
  • 1
    Search tends to cause tons of problems recently. Looks like heavy trolls chose this is a target, and SE running out of creative ideas how to stop them. :/ – Shadow 10 Years Wizard Nov 16 at 11:42
  • 1
    Turns out the previous search bug report was never fixed, so it's duplicate. – Shadow 10 Years Wizard Nov 16 at 12:00
  • 7
    @Shadow10YearsWizard IMO, this should not be considered a duplicate. Questions like this are trouble tickets. Such questions are how problems are reported to Stack Exchange. Closing this question as a duplicate implies that SE doesn't need to know that the problem is back, or that it's much more pronounced now. If a problem goes away and isn't affecting a large number of people for a considerable time (e.g. almost 2 months in this case), but then something similar starts happening, then it should be re-reported, which means posting a new question. – Makyen Nov 16 at 12:10
  • 2
    @Makyen I disagree. Same bug should not get more than one report, or else meta will be flooded with 1000 of the same bug report each time a major bug happens. Exception is when the bug is marked as status-completed, then if it's back, even same bug exactly, new report is the way to go. – Shadow 10 Years Wizard Nov 16 at 12:18
  • 7
    @Shadow10YearsWizard For all you know, the previous bug may have fixed itself before the team got around to responding to it, then this one could be a new bug. I'd let this question continue, and if it is indeed the same bug, SE staff can close it. I've seen that happen many times. – Sonic the K-Day Hedgehog Nov 16 at 12:25
  • 2
    @Shadow10YearsWizard it is too early to call. This bug gets a second chance. – rene Nov 16 at 12:41
  • 2
    @rene Well I do stand behind my closure, but won't (and can't) fight further in case it's reopened. Let's just hope SE developers will also give it so much attention as we all do. :) – Shadow 10 Years Wizard Nov 16 at 12:47
  • 1
    Now the search is working again... what happened?? – mindstormsboi Nov 16 at 15:18
  • 4
    @Makyen again, I disagree. The outcome is 100% the same, and that's a fact: search does not work across all sites. I don't see a point to report it every time it happens while there is still an open and pending bug report exactly for that. As it turns out, SE developer reopened this one without any comment, I still think it's the same bug but I can't really do anything further. – Shadow 10 Years Wizard Nov 16 at 17:34
  • 3
    @Shadow10YearsWizard I understand you disagree. Frankly, at the moment, it doesn't matter if it's the same bug or not (there is no way for us to actually know if it's really a duplicate, without input from SE). What was needed is that it gets communicated to Stack Exchange employees that a problem has suddenly gotten severe, even if it's the same problem. The way that is done is by having an open question. Quickly closing the question as a duplicate of a months old problem report tends to hide that information from SE, which is detrimental to the overall functionality of the network. – Makyen Nov 16 at 17:54
  • 1
    It seems to be functioning okay now. – Ollie Nov 16 at 18:34
  • 2
    +1 to what @makyen wrote. Useful for us to keep active issues open, even if it's recurrence of an older issue (and not necessarily the same root cause) – Yaakov Ellis Nov 16 at 18:57
15

One of the nodes in our primary Elastic cluster has been having a few issues (we suspect a hardware issue at this point) that have resulted in it randomly re-allocating shards across nodes. For, as yet, unknown reasons it drastically degrades cluster performance resulting in timeouts and intermittently taking the entire cluster offline. We're in the process of increasing capacity in that cluster as well as provisioning a replacement search node for the one that keeps blipping.

For the moment we've moved search traffic over to our failover cluster while the capacity increase and replacement nodes happen in the primary cluster. It'll be a bit slower but won't completely break like before :)

Apologies for the inconvenience!

| improve this answer | |
6

For now the search functionality is completely broken, both on MSE and SO (and most likely network wide). Search times out after a while and serves this beautiful page:

enter image description here

| improve this answer | |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .