Forum Discussion
Nimbostratuswhat is it with APM and mqx 64k size of a HTTP body?
hi out there we are running version 11.3 of the os and I have hit a annoying problem with a vs which act as frontend for a file-server where up and download through HTTP.
See here - here is a Little part of the ltm log where a client on 195.82.25.34 connects to file.online.com on 95.8.53.41:
Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_Cookie_Clientless_mode : Clientless-mode Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : ============================================= Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Client 195.82.25.34:5506 -> file.online.com/GMInboundService/GMInboundService.svc (POST request) Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Content-Type: text/xml; charset=utf-8 Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : SOAPAction: "http://tempuri.org/IGMInboundService/SendFile" Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Host: file.online.com Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Content-Length: 81047 Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Expect: 100-continue Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Accept-Encoding: gzip, deflate Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : Connection: Keep-Alive Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : clientless-mode: 1 Aug 19 14:21:07 hsp-f5 info tmm6[13630]: Rule /dk_dmz/DK_log_headers : ============================================= Aug 19 14:21:07 hsp-f5 err tmm6[13630]: 01230140:3: RST sent from 95.8.53.41:443 to 195.82.25.34:5506, [0x171f4d4:2330] APM HTTP body too big Aug 19 14:21:07 hsp-f5 err tmm6[13630]: 01230140:3: RST sent from 95.8.53.41:443 to 195.82.25.34:5506, [0x16c8e24:1305] TCP 3WHS rejected Aug 19 14:21:07 hsp-f5 err tmm6[13630]: 01230140:3: RST sent from 95.8.53.41:443 to 195.82.25.34:5506, [0x16c8e24:1305] TCP 3WHS rejected
I have defined the kernel flag to log TCP rst cause which tells me that the APM HTTP body is to big - if I run without APM the vs Works fine - where it acts as solely reverse proxie - but I need to use the APM to get a certificate based kerberos sso up and run (which also Works fine up to file-sizes of ~ 48k) so I think I am forced to use the apm and get a workaround on this Little problem. Any suggestions on how I can come around this 64k limitation? best regards /thomas iwang
15 Replies
- boneyard
MVP
what does this ENG HF actual do? introduce that extra variable, why wouldn't that be included in the latest normal HFs by now.
on 11.5.1 HF5 it doesn't exist
[root@brt-bigip-01:ModuleNotLicensed:Active:Standalone] config tmsh list sys db tmm.access.maxrequestbodysize 01020036:3: The requested database variable (tmm.access.maxrequestbodysize) was not found. 
RobertWebb_7911boneyard,
Yes, the HF introduces the that variable so you can get past the 64K limit. Why was it not included in 11.5.0 and 11.5.1? No one will say. When I opened this ticket, the original answer was, we do not have a clue why this is happening. They wanted a qkview and all kind of other information. I gave them proof without all that that is was indeed an issue. The case engineer went to his higher level engineers and they all confirmed that this was a limit.
When asked if they were going to fix it, I was told that they may not because I did not provide all the standard BS data they wanted. So when I came across this original posting and was able to find the ID for the fix, I soon found out that is was implemented in 11.3 and 11.4 via a HF and in the baseline for 11.6. However, 11.5.0 and 11.5.1 were both skipped. After I provided all the data on this, it was called the needed, "smoking gun" and was accepted by product development for an HF.
My question is if it was put in 3 other versions, why did engineering not have that information and be able to do something about it from day 1. My guess, is that product development and engineering are so far seperated that there is a major disconnect.
And to your question, it doesn't exist in 11.5.1 HF7 either, which is the latest. Mark posted that he had it in hand and deployed, so I have just sent an email over to my case engineer to find out if he has gotten the HF yet as I am up against a project deployment timeline.
Mark_van_DCirrostratus
Robert,
Am running 11.5.1HF7. Ask your engineer to check C1754531.
- RobertWebb_7911
Mark,
Thanks... Did your engineer end up referencing the case number I gave you? My guess is that the HF came out while my engineer was off and he just hasn't gotten back to me. THat is why I sent a reminder this morning.
- Mark_van_D
Robert,
Am having a few issues after the EHF. May be unrelated but will let you know what support comes back with.
