![]() Want to use generaldelta today is because of adverse performance impact I still think using generaldelta is advised. Will really help (social and technical solutions Known, so we can say with confidence that:Ĭ) preventing commiter's from interrupting a commit I'm going to go out on a limb here,Īnd say we can do an hg server upgrade in Q3.Īs for the "wedged" problems - we have gotten someĭecent debugging (thanks to Glandium) and theories ![]() Like we'll be able to fasttrack upgrades withing a point And, fromĭiscussions with a "major hg user" (facebook), it seems We did get in one server upgrade last year. Largely "no one owned hg" (imo), so resources were notĪligned across the groups in a timely fashion. The issues blocking hg server software upgrade were "wedged hg" problems when resetting was a viable As far as I know that wasĭue to people not spending the time to debug the That resetting has been done before mainly due to I'dĭefer to Taras on that issue too, of course.ĭon't have a good update here. So I'd rather us to keep that in mind before ![]() Underestimating) and it hasn't happened yet, Version on hg.m.o for a number of years now (I We've been talking about upgrading the hg Version of hg, but we should also note that On the issue of the hg repos, I'm of courseįine with running the latest and greatest Help (social and technical solutions available) Much more is known, so we can say with confidence that:Ī) there is no silver bullet server releaseĬ) preventing commiter's from interrupting a commit will really (thanks to Glandium) and theories (thanks to GPS) within the last Go out on a limb here, and say we can do an hg server upgrade in Q3.Īs for the "wedged" problems - we have gotten some decent debugging With a "major hg user" (facebook), it seems like we'll be able toįasttrack upgrades withing a point release of stable. Now Laura owns repos, so "things have changed". Owned hg" (imo), so resources were not aligned across the groups inĪ timely fashion. The issues blocking hg server software upgrade were largely "no one The "wedged hg" problems when resetting was a viable option. As farĪs I know that was due to people not spending the time to debug Has been done before mainly due to respository corruption. I'd defer toĭon't have a good update here. Hasn't happened yet, so I'd rather us to keep that in mindīefore putting all of our eggs in that basket. Version on hg.m.o for a number of years now (I canĬonfidently say 2, but I'm probably underestimating) and it Running the latest and greatest version of hg, but we shouldĪlso note that we've been talking about upgrading the hg > On 5/27/14, 10:22 AM, the issue of the hg repos, I'm of course fine with > On Tue, at 1:48 PM, Gregory Szorc > wrote: > putting all of our eggs in that basket. > hasn't happened yet, so I'd rather us to keep that in mind before > now (I can confidently say 2, but I'm probably underestimating) and it > talking about upgrading the hg version on hg.m.o for a number of years > and greatest version of hg, but we should also note that we've been > On the issue of the hg repos, I'm of course fine with running the latest > any updates on what we want to do here, Taras? Taras previously said that this is on his radar, > changesets to live forever as they would essentially be the data backing > requirement of the hg hosting side that I know of is for the published > Resetting repos is definitely going to be a problem. I know we want a minimum viable product, but I consider some plan for scalability to be part of that minimum putting this off until later will cause more pain in the long run. As gps remarked, the try server has caused a lot of headaches with its thousands of heads, and I'm pretty sure no one in releng wants to go down that path again unless we have a really solid plan. Unless we are treating the initial deployment as a temporary throwaway solution (and I don't think we are), we need to deal with maintenance questions now. What kind of maintenance do we need going forward? I noticed a thread on hg scaling in this group, but I didn't really see any conclusion. Now that fubar is back from PTO, if we need something large, we need to start setting it up now. Does contain everything we need from that side? What kind of disk space do we need? When we set up and around 6 months ago, the idea of having an hg review repository wasn't on the radar, so we probably don't have much disk space available right now. It's pretty far outside my domain, much more of a releng speciality than A-Team. I am, however, still confused by the hg-review-repo side of things. The Bugzilla integration is just about complete we just need to polish the comment-diff mirroring, and we should be good.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |