Tom Whitten wrote: > Sean Wilcox writes: > >> Fixing the code as was talked about for the restart timer bugs, I did >> find that the exit status was being captured in the wait.c code, >> providing a way to track this. And added the code to check if status >> was 0 then reset the timers if not do not reset the timers. >> >> And it seemed to be working nicely but as I was writing up a couple unit >> tests to add to the manual console-login testing I did, I found the >> following in the man page of svc.startd(1M): >> >> "Wait" model services are restarted whenever the child pro- >> cess associated with the service exits. A child process that >> exits is not considered an error for "wait" model services, >> and repeated failures do not lead to a transition to mainte- >> nance state. >> >> Given this statement I have taken out the check of the exit status and >> just reset the start timers in the event of a wait style exit. Just as >> a note this really isn't the case today, as shown by creating a service >> that uses a duration value of child that exits ten times rapidly. >> >> new webrev is located at : >> >> http://cr.opensolaris.org/~swilcox/webrev.1/ >> >> Thanks, >> >> -- >> Sean Wilcox >> 303.272.9711 >> x79711 >> >> _______________________________________________ >> smf-discuss mailing list >> smf-discuss at opensolaris.org >> > > Sean, > > It looks good to me, but I've not spent a lot of time looking at restarter > code. Thus, it would be good to get a second opinion. > > tom > _______________________________________________ > smf-discuss mailing list > smf-discuss at opensolaris.org > > > I have reworked this fix now to take into account not putting hooks into the graph code to tie it to the starter. And would like to request a re-review of the new bits :
http://cr.opensolaris.org/~swilcox/webrev2/ Thanks, -- Sean Wilcox
