Do you have imports and strmlnks? That’s what is causing the slowdown.
The server in my M1 Mac mini slowed down to a crawl with imports. I mean a crawl, deleting tool for ever. Browsing the server pages took a while. Launching any clients took a while.
All clear after I took the imports out, now it’s screaming fast.
Synology DS1019+ 0183a503-b99d-40b9-96fe-43aa2638b36b
Synology DS1513+ ad560681-a504-485d-9018-8338483b7748
Yes, but nothing added to them lately. Has to be something in the DVR code that changed recently.
Take out the imports on one of the Dvrs and test. It fixed my slowness.
The Mac m1 can comskip a half hour show in 34 seconds, a two hour show in 4 minutes but if you have imports, you will die waiting for something load, delete or navigate.
I'm seeing the web UI is making a lot of repeated requests to the backend for some reason, causing the high cpu usage and lagginess.
Which screen of the web UI are you on? Does it make a difference?
This doesn't really make any sense. Maybe if you had tens of thousands of imports (i.e. if you pulled in the entire streamlinks repository).
No, it doesn’t make sense but it’s true. 14k tv show episodes, not unreasonable if your importing your library, no streamlinks at all.
I've got Import directories setup on both but I'm just testing small samples.
One NAS has 0 Imported recordings, the other has 173.
I'm 'tailing' the Status Events from both NAS's (curl -Ns -m 7200 http://x:8089/dvr/events/subscribe)
and don't see anything unusual when the CPU usage spikes. Actually I'm passing the curl output through jq and redirecting output to a log file, then tailing the log file with Notepad++.
I don't see anything in the DVR log or events when this happens.
Since I originally posted it's happened 3 times on one NAS.
There is corresponding network send activity happening at the same time as the CPU usage spikes.
With http logging enabled can you see which requests come in during that time and from which IP.
Enabled. If I send diagnostics right after it happens will you see the http logging?
Just happened again, 135db9e5-5bf0-4c99-97b3-ea6790952658
About 17:08:50 local time
The requests are coming from 192.168.1.2, presumably via an open browser window on that machine.
Yes, that's my Windows PC using Firefox to access the web UI.
I just closed that web browser tab at 5:30pm and won't reopen it until recordings are done tonight.
Not sure why so many requests, most of the time the browser was sitting at the main settings screen.
Aman, if you look at the http events GET "/dvr/events/subscribe"
Do the times seem off to you?
i.e. 1h59m59s was me logging for 2hrs, but the rest?
2021/02/14 16:47:07.813671 [HTTP] | 200 | 48.823545515s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:47:21.785961 [HTTP] | 200 | 13.503223743s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:48:47.653107 [HTTP] | 200 | 39.346589286s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:49:08.515908 [HTTP] | 200 | 1h59m59s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:51:29.726655 [HTTP] | 200 | 34.555734096s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:54:44.774886 [HTTP] | 200 | 49.261561201s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:56:50.857740 [HTTP] | 200 | 2.825011012s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 16:59:48.360558 [HTTP] | 200 | 4.108589354s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:00:16.259811 [HTTP] | 200 | 7.821580517s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:04:35.146378 [HTTP] | 200 | 21.549877471s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:05:20.436670 [HTTP] | 200 | 23.693474577s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:05:54.280918 [HTTP] | 200 | 28.455956028s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:08:42.224751 [HTTP] | 200 | 1.065664415s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:10:09.632768 [HTTP] | 200 | 47.361539645s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:14:39.505566 [HTTP] | 200 | 2m41s | 192.168.1.2 | GET "/dvr/events/subscribe"
2021/02/14 17:27:42.565740 [HTTP] | 200 | 1m3s | 192.168.1.2 | GET "/dvr/events/subscribe"
Just wanted to add here that when I was trying to troubleshoot my issue with the Quatro, I also tried the new DVR beta and had the same issue. IO Wait spikes and causes load numbers to spike too.
I'm not sure what causes it, but it went away after I switched back to
2021.02.04.1931
Good to know I'm not alone.
Appears that something in the newer versions causes the web UI to get too 'chatty'.
Not sure what's causing the CPU spikes, but since I closed the browser tab on it about 3hrs ago the CPU usage looks normal. Last spike was right before I closed the browser tab.
What page was that browser window open to?
I did move around some between web UI tabs yesterday,
Trying again today since no recordings scheduled.
At 9am I enabled html logging and updated to pre-release 2021.02.15.1636.
I'm going to leave the web UI open on the settings tab and not mess with it.
Will see what CPU usage looks like.
Have already seen 10 mins of high CPU usage due to guide update this morning, but nothing since then.
