Bug ID 696468: Active compression requests can become starved from too many queued requests.

Last Modified: Oct 01, 2018

Bug Tracker

Affected Product:  See more info
BIG-IP APM, GTM, LTM, vCMP(all modules)

Known Affected Versions:
12.1.0, 12.1.0 HF1, 12.1.0 HF2, 12.1.1, 12.1.1 HF1, 12.1.1 HF2, 12.1.2, 12.1.2 HF1, 12.1.2 HF2, 12.1.3

Fixed In:
14.0.0, 12.1.3.1

Opened: Nov 28, 2017
Severity: 2-Critical

Symptoms

From the "tmctl compress" table: the cur_ctx value for QAT is equal or higher than 512, and the cur_active remains at zero. CPU utilization per tmm in this condition may be quite high.

Impact

Compression on a per-tmm basis can stop servicing new requests.

Conditions

At least 512 contexts with no traffic wait in the compression queue and prevent new requests from getting compression service.

Workaround

Switch to software compression.

Fix Information

Soften the restriction so that accumulated contexts with no traffic cannot not prevent busy contexts from getting compression time.

Behavior Change