Today, Stack Overflow went read-only in the morning (IST) and came back after some time.

enter image description here

Now again it went read-only.

enter image description here

This pertains only to a specific site but the announcements were made on @StackStatus and on stackstatus.net and it is common for the entire network.

Both of them are updated on 21st July and no new information after that. So, the question is:

If the maintenance is for only a specific site, won't it be updated on the above two resources?

  • 4
    Cross site duplication Why is Stack Overflow in read-only mode? – Arulkumar Jul 24 '17 at 14:15
  • @Arulkumar no my question is the bolded text at the bottom(won't the information be added to resources) + that post is made for the morning read-only – Optimus Prime Jul 24 '17 at 14:16
  • 1
    Something not good is going on somewhere in SE, maybe hosting, maybe database.... MSE now suffer from occasional 404 errors as well. Hopefully, they will fix it soon. – Shadow Wizard Wearing Mask V2 Jul 24 '17 at 14:17
  • only happens to SO. not even MSO.(made me think that they are separating SO from other SE sites) – Optimus Prime Jul 24 '17 at 14:18
  • 1
    This was not scheduled - people are looking into the causes. – Oded Jul 24 '17 at 14:19
  • 2
    "made me think that they are separating SO from other SE sites", that has been the case for a long time already. – Patrick Hofman Jul 24 '17 at 14:27

This is not planned.

There's an issue with connection pool timeouts I'm investigating. Normally, we disable said spill and opt for a few error pages when it happens over a read-only self-recovery, but due to the dynamics of the data center failover we had this re-enabled. It's just been disabled in a prod build that just went out and I'll be digging into the connection pool issues now.

There are a lot of contributing factors that can funnel into this result. It's ultimately the result of us being efficient here. While that's often a good thing (fewer servers to maintain, buy, etc.), it's not so good for per-server bottlenecks like connection pools. We can push those to the limits fairly routinely. Today, we're pushing way past those limits and I'm currently digging into why.


You must log in to answer this question.

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