Hi all,
We've discovered today that there's a default lock timeout of 15 minutes when using Star.
It can be overridden by "set lockmode...".
I think it's set in rqfcall() (rqfmain.c).
Curious as to why* the usual session default timeout is overridden for Star sessions.
Any ideas?
Geraint
* Emboldened at the request of a colleague who shall remain nameless but is known to become agitated whenever the subject of Star's features/curiosities/peculiarities is raised.
Well, it's a message reply timeout, not a lock timeout. I might have set it higher if I were doing it, but some sort of timeout is needed in case the server at the other end dies or becomes inaccessible.
Well, it's a message reply timeout, not a lock timeout. I might have set it higher if I were doing it, but some sort of timeout is needed in case the server at the other end dies or becomes inaccessible.
Two things then:
1. Why is set lockmode session timeout affecting a message reply timeout?
2. Shouldn't the Star server keep a heartbeat thread going to all the other servers involved and leave the querys alone?
PS. STAR gets me so riled up I may burst into song. I have a new version of Delilah that'll knock your socks off.
I was on a conference call about an hour ago in which just that dreadful prospect was invoked as a warning to us all.
Also you have misspelt "rot" as k-n-o-c-k.
PS: Star riles me up too. We had a site where we ran it with an Oracle gateway. It was soooooo close to being an elegant solution too.
GJ has just told me that 'rot' implies a slow process of putresence, whereas my singing is far faster than that.
STAR is so close to being very useful, but its limitations keep getting in the way. I suspect the trouble is that its trying to be very clever and handle too many possibilites. What I'd like to see is a reworked 'STARlite' that was just talking Ingres to Ingres and which sat within a normal DBMS server.
Login
Register
Actian.com
15 minute lock timeout in Star
Linear Mode
