Everything was going smooth until this evening. The server started responding very slowly so I decided to stop the service and install the latest update and reboot. Now it’s once again being extremely sluggish. I downloaded the logs once the web UI became available again.
After updating to 4288 and the next version after I gave the server an hour before I allowed connections again.
It must still be working on the database content migration (updating) which was discussed.
It was supposed to start with PMS build 4200. There were issues with that build.
4288 is with the runtime library issues fixed.
We got one last issue yesterday with 4264.
Database upgrading is a single-threaded task.
Your CPU, even though it has a blazing 14,000 passmarks:
Is a 12 core.
The net result is , in a single-threaded task, it’s going to be as-fast or even slower than other systems.
My quad core i7-7700 is 8700 passmarks.
This CPU can do the database work twice as fast as that Xeon.
Xeon: 14,785 / 12 = 1,232 passmarks per core
Core i7: 8612 / 4 = 2153
That said, if we had identical databases, a core i7-7700 would finish in 1/2 the time.
How much media is indexed?
AND
when PMS is completely idle (no playback) or other activity,
Is the CPU busy or idle ?
On that CPU, top (not htop) will show if a single core if maxed. you’ll see a '100` somewhere -OR- you’ll see total loading at just over 8.33%
You logs are also showing it’s scanning for media.
Didn’t you ever set it for partial and automatic ?
I’ll have to check the CPU with top while the server is mostly idle to see. I’ve just been relying on when the logs go idle.
I’ve optimized my database since today’s update if that will solve anything.
I still have my server set to only scan partially when changes are detected since you mentioned last time.
Is there any easy way to tell how much I have indexed?
I stopped Plex, and rolled it back to 4163 and am updating back to 4264. I see the CPU going between 50-100% so once it’s idle for a few minutes consecutively I’ll re-enable remote access.
Ok, I’ve waited until the web UI is available again, and the cpu is mostly idle. Opening the flood gates.
How can we verify whether or not these database migrations have completed?
Use top on Linux (not htop). CPU usage should return to normal.
Of particular note: you should not see any one CPU core at/near maximum.
I/O should be idle (no more database i/o occurring).
e.g. top should not show any cores more active than the others.
NAS users usually have a graphical meter. 25% of a quad core = 1 core maxed, etc.
You know when the CPU isn’t sitting idle, right?
Yes, I was hoping for something in the logs or something. Are these migrations occurring with every update as of late? It seems like some users are reporting the migrations occurring every time they update or is that because they’ve been bouncing back and forth between versions? I have a script that runs every morning and updates if one is available so I’m assuming they should have been completed with the first update that installed where those migrations needed to occur and every one since should have updated without the migration. But some of the other posts in this thread have me questioning that.
I’d like to point out, the release notes for the latest versions have been nearly identical (except 4228), and neither mention this crucial database change.
Plex Media Server 1.22.1.4228 is now available to everyone.
Plex Media Server 1.22.2.4256 is now available to Plex Pass users in the Beta update channel.
NEW:
FIXES:
Plex Media Server 1.22.2.4262 is now available to Plex Pass users in the Beta update channel.
NEW:
FIXES:
The only difference being “* (Collections) Collection deletion would fail”
A change that is causing this many issues for people should really have a disclaimer clearly stating it could take some time on first boot and to be patient, and to explicitly state somewhere that there is a database change - and which version is the one actually making the changes.
It was supposed to start with PMS build 4200. There were issues with that build.
4288 is with the runtime library issues fixed.
We got one last issue yesterday with 4264.
Database upgrading is a single-threaded task.
^^^^^^^^^^^^^^ This should have been in the release notes.
After installed Version 1.22.1.4228 I can not reach library from any devices,i pad,apple tv,android devices,I restarted server and re-installed all app in devices but it is not working, I can not reach my library after update new server version.
Please note that due to some particularly large database migrations, your first time starting up the server with this release may take an unusually long time, depending on library size, disk speed, and CPU performance.
Is this what you were saying wasn’t in the release notes?
I provided the finite technical details. Not a lot of people understand what “single threaded task” means.
Can we get some kind of progress indicator in the logs, at least, that let us know what the heck is going on with the database migration. Simply having it stop working for the duration of a process we have no way of knowing is happening and are given no clues about progress is terrible.
I haven’t been able to access my server all day, it was working last night… log files periodically are updating with “Unable to get Gateway” messages, so I know that the process isn’t actually dead but it won’t respond to any requests from clients or serve the web client. How can I be sure it’s doing something useful? It doesn’t make sense because the previous communication was that database migration occurs on “first time starting up the server”, so why is it when I kill the server and start it again it works fine and then dies overnight into this state?
Yes, I hadn’t scrolled up enough to see 2 posts for 4228 - that’s my ignorance. I do not recall seeing that disclaimer in the release notes presented on the server when the update was available though.
By the time we get a progress indicator added, this even will be OBE,
I will suggest to the Engineering they consider an indicator for future tasks.
A new update was released today. I went ahead and applied it, so I hope it resolves our issues.
1.22.2.4274-a85c82272
Can you advise if the database changes affect this method of changing media locations
or are those still the only 4 tables that need updated?
Latest update (public) has officially crashed again. I don’t understand why after 4+ versions including beta, this is still not resolved. I am going back to 4163 and not moving. I give up honestly.
A problem in the build system caused today’s train wreck.
We’ve got it back.
I’ve tested on Debian and it’s all working.
PMS 1.22.1.4275 is deploying to Public now.
PMS 122.2.4276 (if it passes testing) will deploy to Beta
PMS 1.22.1.4275-48e10484b
Deployed. Reports are it’s working again 
mine just crashed, don’t get your hopes up. Here’s the logs, but I doubt it’ll help since nothing else has.
Plex Media Server.2.zip (976.1 KB)
It is working after update PMS 1.22.1.4275 version. Re start server and looks ok
