Bad gateway 502

[gnif]

FWIW I had assumed this occasional 502 was a cloudflare issue. Even though the message specifically explains there is no response from the server, it is proudly covered in their branding.

CloudFlare is just reflecting the 502 that the server is sending, if it was only a CloudFlare issue I would not see the 502 responses in the server's logs.

I have no issue at all with people reporting information, in fact it is encouraged, but unless someone actually sends me a PM, I don't always see or know if there is an issue (In this instance I stumbled across this thread).

Along with the info other incidences I've had could involve any individual thread and in some cases a whole board.

It really depends on what has been locked, there are tons of different updates that occur across the database all the time, even thread read counts.

Sounds like Dave's in for a server upgrade again.

Time for a couple SSDs?

This only makes sense if there is a HDD related bottleneck, this has not been established yet, nor do the current metrics point to such an issue.

I can think of a couple of reasons.  Your initial explanation was not consistent with observed symptoms.

How so? I stated the server is experiencing a load issue, this has not changed.

Your force/refresh (Ctrl+F5) or clearing cookies/cache is only working by coincidence, its just the fact that you made another request to the server when it was a little less loaded and as such it was able to return a result.

[David Hess]

I can think of a couple of reasons.  Your initial explanation was not consistent with observed symptoms.

How so? I stated the server is experiencing a load issue, this has not changed.

If the problem is a load issue, then improving performance will rectify it.  If a limited number of processing threads are getting stuck because database rows are stuck in a locked condition, then the problem is why they are locked or why they were not unlocked.

Or are you suggesting that specific database rows get locked for hours at a time and this is just because of insufficient server capacity?  It seems odd that such a state would persist for such a long time.

[tautech]

Or are you suggesting that specific database rows get locked for hours at a time and this is just because of insufficient server capacity?  It seems odd that such a state would persist for such a long time.

This.

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

I won't challenge gnif on any of this, only reporting observed behaviour.
I might add at times my connection speeds are painfully slow (10's of Kb/s), even at full speed my DSL is only 1.7 Mb/s. 
Tethered to my phone it's heaps better however if I'm getting errors already, nothing changes.

[David Hess]

Or are you suggesting that specific database rows get locked for hours at a time and this is just because of insufficient server capacity?  It seems odd that such a state would persist for such a long time.

This.

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

I am assuming there is more than one problem which returns the 502 gateway error.

In the past I got that error for days at a time and clearing cookies and cache fixed it; it would be awfully coincidental if after days, I just happened to clear cookies and cache at the same time that the problem was resolved on the server side.

The current error and most of them in the past were not affected by clearing cookies and cache.

And then there is a third thing going on where Cloudflare issues the redirect destroying the browser history so you cannot recover your post or reply.

[tautech]

Or are you suggesting that specific database rows get locked for hours at a time and this is just because of insufficient server capacity?  It seems odd that such a state would persist for such a long time.

This.

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

I am assuming there is more than one problem which returns the 502 gateway error.

In the past I got that error for days at a time and clearing cookies and cache fixed it; it would be awfully coincidental if after days, I just happened to clear cookies and cache at the same time that the problem was resolved on the server side.

The current error and most of them in the past were not affected by clearing cookies and cache.

And then there is a third thing going on where Cloudflare issues the redirect destroying the browser history so you cannot recover your post or reply.

I may have had that once IIRC but Chrome pretty much plays nice with the forum.
I get 2 gateway fails, Chrome reports 502 (most common) and Cloudflare much less often.

[StillTrying]

502s all over the place 5 mins ago, noticed site slowing just before the 502s.

Didn't do any clearing of cookies or caches - never have,  just waited 2 mins.

[Iwanushka]

looks like we have to many users especially when entire US/AUS/NZ jumps in maybe it's time for nginx and load balancing if Dave has the money for multiple Servers/DB clustering

[james_s]

I frequently get the Bad Gateway error, usually I can resolve it by clearing my browser cache. I tried this after noticing when it starts happening in one browser I can load the site just fine in a different browser on the same PC.

[StillTrying]

Right. Page 1 is stuck on 502, but page 2 is fine of:
https://www.eevblog.com/forum/beginners/change-of-capacitor-capacitance-in-a-circuit/

Later...
I used ALL in the end to see pages 1+2.

[dimkasta]

Gnif may I ask what maintenance operation locks the db for so long? I don't know if you want to reveal that info publicly though.

You could try increasing the nginx timeout (duh) temporarily until you put your finger on the db issue.
Also, have a look in case there is any forgotten MyISAM causing table locks somewhere that it shouldn't.
And you could fiddle with innodb io capacity if you see any issues there and you can spare disk i/o.

An operation taking so long on rows rather points to some kind of i/o bottleneck. Are you using a swap file ?

[tautech]

Had 502's again a few minutes ago, right on midday NZ time, 9am Aussie.
Net was slow so switched to a USB tethered cell for service....problem gone. 

Just letting you know so you can check logs.

[StillTrying]

https://www.eevblog.com/forum/login2/

Was stuck on 502, for a couple of minuettes.

[tautech]

Still getting errors for https://www.eevblog.com/forum/unreadreplies/

But https://www.eevblog.com/forum/unreadreplies works. 

[xrunner]

I get it a lot lately too but I never say anything here because the message states that the main players know about the problem. 

[tooki]

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

No, it doesn't. What fixes it is waiting a few moments and trying again — which is exactly what is happening when people go to their settings, clear the cache, etc. It's the time delay that's fixing the problem, not the cache clearing!!

[tautech]

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

No, it doesn't. What fixes it is waiting a few moments and trying again — which is exactly what is happening when people go to their settings, clear the cache, etc. It's the time delay that's fixing the problem, not the cache clearing!!

I'll stand by what I have said. 

And this won't work ATM either: https://www.eevblog.com/forum/unreadreplies/?start=0

The forum has been trouble free for a few days.....until this morning. 

All we can do is relay as much info as possible to gnif while he tries to get to the bottom of it.

[tooki]

Not just hours, days in some cases and a browser cleanup always fixes it.....until the next time it happens.

No, it doesn't. What fixes it is waiting a few moments and trying again — which is exactly what is happening when people go to their settings, clear the cache, etc. It's the time delay that's fixing the problem, not the cache clearing!!

I'll stand by what I have said. 

And this won't work ATM either: https://www.eevblog.com/forum/unreadreplies/?start=0

The forum has been trouble free for a few days.....until this morning. 

All we can do is relay as much info as possible to gnif while he tries to get to the bottom of it.

You made an absolute statement, and as such, a single point of conflicting data disproves the claim. I've never seen a forced refresh fix it, ergo, the claim is disproven.

Regardless, the fact that the unread replies page isn't currently working for you, but is working for me, lends credence to it being a server-side issue involving specific record locks, as gnif has explained.

[StillTrying]

502 on
https://www.eevblog.com/forum/beginners/signal-changes-when-second-probe-is-attached-to-op-amp/

getting
https://www.eevblog.com/forum/beginners/signal-changes-when-second-probe-is-attached-to-op-amp/msg1167650/?topicseen

from history got me the page.