On 11/11/20 6:46 AM, roie rachamim wrote: > The default flow is REQ_ICAP1->REQ_ICAP2->RESP_ICAP1->RESP_ICAP2 > using the dynamic routing i'm able to skip REQ_ICAP2 (by setting > X-Next-Services to 'RESP_ICAP1, RESP_ICAP2') > I'm also able to skip RESP_ICAP1 (by setting X-Next-Services to RESP_ICAP2') > What i didn't manage to do is to stop adaption completely by returning > an empty value for X-Next-Services > According to the documentation: "An empty X-Next-Services value results > in an empty plan which ends the current adaptation." > So i'm not sure what am i missing. Maybe with empty value it simply > skips to the response chain? If that is what happens, then it is probably a Squid bug. AFAICT, an empty plan returned by a routing REQMOD service should prevent RESPMOD adaptation by design -- a routing service is supposed to have full routing control. I checked the code that processes X-Next-Services and did not notice any red flags, but the bug may be located deeper in the code, where Squid reacts to the remaining adaptation plan. I suspect somewhere an empty plan is misinterpreted as no plan. If the bug is confirmed, we may have to preserve it for backward compatibility sake and add two special X-Next-Services values (e.g., "END_ADAPTATION" and "JUMP_TO_RESPMOD") to allow updated services to distinguish the two use cases instead of returning an empty list with an arguably ambiguous meaning. > Should i use the adaptations in a different configuration? Ideally, the bug should be confirmed and fixed. If you cannot (facilitate a) fix, then you may be able to work around the bug using service-set annotations and adaptation_access directives (for all or just the buggy cases). Unfortunately, I do not remember whether Squid supports setting transaction annotations using ICAP services. IIRC, Squid supports that for eCAP services. https://wiki.squid-cache.org/SquidFaq/AboutSquid#How_to_add_a_new_Squid_feature.2C_enhance.2C_of_fix_something.3F HTH, Alex. _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
