Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
HristoGrigorov

R77.20.85 performance issue on centrally managed SMB

Guys,

That build is causing significant traffic delays and CPU load is higher than that of R77.20.81. 

Any of you experiencing similar problem ?

123 Replies
Naftali_Oziel
Collaborator

Hi, which appliance are you running this on 700/900/1400.  Plus which blades to you have enabled?

0 Kudos
HristoGrigorov

Centrally managed 1470 appliance. FW, VPN, AC/UF, IPS blades. Same configuration is fine on R77.20.81. 

Currently one node of the cluster is running 81 and one 85. I flip them over from time to time and as soon as 85 becomes active load average increases well above 1.0 while on 81 it rarely goes up to 0.8.

0 Kudos
Naftali_Oziel
Collaborator

Ok. Thanks.  Do you have any standalone (SMB)  and those to SMP but in a non-cluster setup?   Checking my end to compare stats but am running 1490 using the same blade setup.   However, are local managed none central or cluster.  

0 Kudos
HristoGrigorov

Sorry, only those two 1470s in a cluster. I don't know why, but that build makes things run really slow. I am going to revert it back to 81 today which is as you know long and painful procedure because flashing previous firmware version is not supported. 

0 Kudos
Pedro_Espindola
Advisor

I have the same problem.

Centrally managed 1470. Build 990172731

Before upgrade, average CPU load was below 20% during work hours. After upgrade, average was at 35% on sunday and now, during work hours, it is peaking at 85%... I can't tell the average, because the gw stops responding to SNMP every few minutes.

HristoGrigorov

Yesterday, I gave it another try. And users immediately started to complain about network being slow. Switched over to the other member (77.20.81) and everything went back to normal. Btw, 77.20.81 build 541 has exactly the same problem. R77.20.81 build 525 is rock stable and performance is noticeably better. So I am staying with that for the time being. 

0 Kudos
Pedro_Espindola
Advisor

Noticed some instability periods in 541 for centrally managed appliances, but 77.20.85 was way worse and instability seemed to be permanent, not momentary. After reverting to 77.20.81 541 things definitely improved!

Had no issues with 541 in locally managed ones. I didn't test 77.20.85 GA on locally managed appliances.

0 Kudos
HristoGrigorov

R77.20.85:

R77.20.81:

Time frame is: 08:30 AM - 09:30 AM (1 hour) on work days. Same policy.

0 Kudos
Pedro_Espindola
Advisor

Did you open tickets with TAC about this build? And what about Build 541? Did they say anything?

0 Kudos
HristoGrigorov

I am sorry, I don't have time for that at the moment. I am tackling another SR with TAC at the moment. I just reverted back so not actually using it at the moment.

G_W_Albrecht
Legend
Legend

Strange - one of our customers has reported performance gains after changing from R77.20.81 to R77.20.85 on a single SMB GW.

CCSE CCTE CCSM SMB Specialist
0 Kudos
HristoGrigorov

That's interesting. Is it locally or centrally managed ? 

0 Kudos
G_W_Albrecht
Legend
Legend

That is a locally managed SMB unit, so there is much difference to your environment.

CCSE CCTE CCSM SMB Specialist
0 Kudos
HristoGrigorov

Yes, it seems issue is only on centrally managed SMBs.

0 Kudos
Naftali_Oziel
Collaborator

locally managed is also running into issues  from a GUI performance getting timeout error msg.   As for actual traffic don't believe there is an impact.

0 Kudos
HristoGrigorov

Thanx for sharing. The poor GUI performance was one of the reasons I went for central management. I am so far very satisfied with 81. It performs beautifully here even with HTTPS inspection enabled. I checked again 85 release notes and there is nothing interesting for me that was changed. I'd love to see some features from R80.20 incorporated into SMB firmware but that is not likely to happen.

Once we solve together with R&D that nasty SecureXL problem with static routes I will be very happy with what SMB is Smiley Happy 

0 Kudos
G_W_Albrecht
Legend
Legend

Yes, it is not recommended to keep a GUI connection open without doing any configuration changes. WebGUI is not really a tool for monitoring as it uses too many ressources on a unit that does not have much...

CCSE CCTE CCSM SMB Specialist
0 Kudos
Naftali_Oziel
Collaborator

GUI has latency/timeout issues even on configuration.  Believe this is all related to the new features added and simply will require optimization.  This issue occurred a few revisions back, can't recall which one.  Hopefully another build will be released to address all these shortly as I'd like to upgrade to production system. 

0 Kudos
Kaloyan_Kirchev
Contributor

Hi,

A client also has locally managed 1430 with R77.20.86 but performance is....low.

I had CPinfo here and looking at it with DiagnoseView but cannot find any solution.

Interesting is that I see that memory is 90% and most used proccess is "fw sfwd"(as I saw this is main 1430 proccess".

Any Ideas? Any suggestions? 

0 Kudos
KAPIL_RANA
Explorer

I have the same problem. 77.20.80 ruining well  but after updating 77.20.81 or 77.20.85  hang networ and

cpu   load  90 or 100 %

Before update  cpu load 20 %

 

Locally  managed 1470

0 Kudos
Naftali_Oziel
Collaborator

Suggest you open SR asap.  Never had issues with r77.20.81 on CPU load. 

0 Kudos
HristoGrigorov

Don't think CheckPoint is going to do something about it until someone opens SR. 

0 Kudos
Pedro_Espindola
Advisor

Yeah, but there is no way I can keep a production gateway in that version while R&D investigates.

Best I can do is try to replicate in a lab and then reach TAC.

I have never seen such an unstable build. IMO this should still be in EA or internal testing.

0 Kudos
HristoGrigorov

Yep, same here. Only that I do not have test lab at my disposal. So, not much of a help really. 

0 Kudos
Naftali_Oziel
Collaborator

Fully understandable not impacting production and agree that this build is not GA ready.  I would open an SR of the issue at hand providing as much info as possible and CP to than attempt and replicate the issue in their lab.  This is what I've done in the past by upgrading giving the logs/cpinfo snap shots, processes etc. and than revert back.  This way they have a way to run with it.   Just my two cents.

0 Kudos
G_W_Albrecht
Legend
Legend

My first suggestion when encountering an issue that does not appear at other customers is a fresh firmware install using USB medium.

CCSE CCTE CCSM SMB Specialist
0 Kudos
KAPIL_RANA
Explorer

I have already install fresh firmware 77.20.80 using medium USB but after upgrade to 77.20.85 facing same problem cpu load 80 to 90 %

0 Kudos
Pedro_Espindola
Advisor

It does appear in other costumers. We have at least 4 different cases here in this topic with the same behavior.

0 Kudos
G_W_Albrecht
Legend
Legend

I have also seen that a WebGUI firmware update made the unit unstable, and only USB firmware install resolved that. But of course, if the firmware is buggy this will not help!

CCSE CCTE CCSM SMB Specialist
0 Kudos

Leaderboard

Epsum factorial non deposit quid pro quo hic escorol.

Upcoming Events

    CheckMates Events