2.4.5 RAM Issues
-
Hello! I recently upgraded to 2.4.5 in my lab environment and the reaction I got in that environment has me terrified to roll it out anywhere. I've been recommend my clients hold off on upgrading. In my lab I have 3 firewalls and two are having issues since upgrading.
The first is a VM in Hyper-V 2019. Since the 2.4.5 upgrade it's RAM maxes out after about 5 minutes of uptime. It then runs really slow but does continue to run. Rebooting doesn't help. I ended up having to rebuild and restore from backup in 2.4.4p3 because it was almost unusable. I tried this VM with 1, 2, and 4GB of RAM both dynamic and static with no change. I see that 11.3 isn't officially supported yet but it is in Azure so I'm not sure what the plan is here.
The other issue I had was with a physical install on a Dell R210 II in HA. Since the 2.4.5 upgrade, the master firewall has been rebooting every 24 hours almost perfectly. I noticed that before it reboots that the command line shows "swap_pager_getswapspace(32): failed" over and over. When it does this, the firewall continues to route but the console and webUI are completely locked up until the firewall is rebooted. This system have 4GB of DDR3 ECC RAM.
It seems to be the 11.3 jump in pfSense might have been a bit premature?
-
@MeCJay12 Sounds like the problems reported here. Lots of people with pfSense virtualised having problems with 2.4.5
I found massive increase in memory and CPU usage with 2.4.5.
-
Interesting. Any idea how that would relate to my physical machine though?
-
@MeCJay12 said in 2.4.5 RAM Issues:
I recently upgraded to 2.4.5 in my lab environment
Curious if you were running snapshots on any of that equipment before the release.. edit- actually scratch that. My SG2220 box below was not running snaps and the Lanner board here was. Ends that theory for me.
Im trying to compare the differences between my firewalls and those of others running 2.4.5 release. I have pretty much the opposite results here so far. The only difference between my Lanner boards and an SG2220 I have is that the 2220 shows "wire" on the memory graph which did increase slightly. The Lanner boards do not utilize it. But you can see the moment I pushed the "go" button to upgrade.
-
No I don't run snapshots. I also wanted to grab my up time graph from my physical machine to illustrate the issue. My RAM and CPU on that machine stayed largely the same its just that the swap space it spazing out now. I would share graphs from the virtual but the system slowed so badly I wasn't able to poll it until after I rolled back.
-
@MeCJay12 said in 2.4.5 RAM Issues:
I noticed that before it reboots that the command line shows "swap_pager_getswapspace(32): failed" over and over
I installed 2.4.5 several weeks ago and today I had the same swap space error and had to reboot. The reboot went unusually slow when starting the packages.