Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
Hristo_Grigorov
Participant
Jump to solution

dmesg buffer flood

Hi,

I noticed dmesg kernel buffer on a 1595 appliance running R82.00.11 (998002800) is being constantly flooded with messages:

[433997.048200] [SIM4];reorder_after_decrypt: Could not get QoS config for ifnum -1
[433997.048411] [SIM4];reorder_after_decrypt: Could not get QoS config for ifnum -1
[433997.064086] [SIM4];reorder_after_decrypt: Could not get QoS config for ifnum -1
[433997.112309]
[433997.112309] FW-1: stopping debug messages for the next 57 seconds. To disable this suppression see sk74580

While it does not bother me, I am sure it does bother the device itself. Perhaps logging it once is enough.

0 Kudos
1 Solution

Accepted Solutions
Dafna
Employee
Employee

Thanks for the feedback. The issue was already fixed and it is part of the next release

View solution in original post

10 Replies
PhoneBoy
Admin
Admin

Anytime you're generating messages, be it to a log file or the terminal, it creates non-zero load.
The fact we're suppressing the message when we see a number of them in a short period of time definitely helps that.
If this is happening often, then it might warrant a TAC case. 

0 Kudos
Hristo_Grigorov
Participant

I don't know what work is being done before generating this log message and I leave who is responsible for this to decide if it is worth fixing. May contact me as well of course.

Endless loops are never good. At the very least they interrupt CPUs, cause re-scheduling of threads, etc.

0 Kudos
HristoGrigorov
MVP Gold
MVP Gold

LoL, there are two of us now 🤣

I noticed the following rather incorrect statement that might be the reason for this log message. I do have configured Internet connection but I haven't enabled QoS on it yet.

0 Kudos
LINnik
Explorer

Hi Hristo,

Just adding data points from our side: We ran into the exact same log spam on all appliances updated to R82.00.11 (Build 823) following the PHP2Shell patching.

After thorough testing, we can confirm this appears to be purely cosmetic—we observed no measurable performance impact or network degradation.

That said, kernel buffer floods are never ideal. Looking at reorder_after_decrypt: Could not get QoS config for ifnum -1, it seems SecureXL (SIM) is attempting to query QoS structures post-decryption for an unresolved interface (ifnum -1), even when QoS isn't active on the link.

Does anyone know if Check Point / TAC has already acknowledged this as a general bug in this build, or if an active case ID exists to track a fix?

Hristo_Grigorov
Participant

Hello LINnik,

Thanx for sharing. I haven't observed any performance issues as well. It messes the output of dmesg and fills in the sfwd.elg file real quick and that's it. 

I can't open TAC case but if you can and you do please keep us up-to-date how it goes.

0 Kudos
HerbertP
Participant

after applying the workaround and after the fixed image for the php2shell i now get the follwing message when scanning for update: 

 
Flash storage life expectancy is low
The flash storage is being written to at a high rate, reducing its life expectancy to less than a year.
 
Is this related to log flooding. How to analyse. 
0 Kudos
Hristo_Grigorov
Participant

Kernel message buffer is stored in memory. No problem with that. I noticed they are also logged to a file but because of the suppression mechanism it is not in a large enough amount to burn the flash storage. Last I checked it was a NAND memory and that can take quite a lot of beating.

Theoretically you can monitor I/O usage in CPVIEW.I/O but not sure how accurate it is there.

Drop in expert shell and run bellow command to monitor writes:

# iostat -d 2

Unfortunately simple diag tools like 'lsof' and 'iotop' are not part of the SMB Linux OS to identify which PID is doing the most write operations. 

If appliance is relatively new and in support than it probably makes sense to open TAC case.

0 Kudos
LINnik
Explorer

Worn-out eMMC has already caused us quite a bit of trouble with the 1550 and 1590 appliances, so I wanted to check and rule that out right away as well.

The iostat -d 2 output looks unsuspicious.

A recursive search shows that the error messages only appear in boot_log.elg.

As soon as I get feedback from our System Architect, I'll be happy to follow up here.

0 Kudos
HristoGrigorov
MVP Gold
MVP Gold

I got this very same warning on an old 15xx appliance while doing pre-upgrade check. Perhaps to mitigate the problem a little bit you could redirect security logs to SD card.

0 Kudos
Dafna
Employee
Employee

Thanks for the feedback. The issue was already fixed and it is part of the next release

Leaderboard

Epsum factorial non deposit quid pro quo hic escorol.

Upcoming Events

    CheckMates Events