That's really great to hear! re EXT4 losing data, you shouldn't lose any data if you use ordered-data log mode. I have had a couple of failures on ext3/4 where the metadata was logged but the data wasn't, so the file was "completed" but data inside was zeros after recovery. Just thought I'd mention that.
On Wed, 29 Nov 2023 at 01:05, Henry Wertz <[email protected]> wrote: > I mean, not automated but I make HEAVY use of the local backend so I can > guarantee you that one is pretty solid. (Including forgetting to unmount > it a few times when I shut the computer off, a couple power outages (or > letting battery run dead in the case of a laptop), I have one on a portable > USB drive where I had the USB cable start popping out now and then for a > while (although I got a "tighter" cable and it's fine now), and even one > disk where I started getting bad sectors. I can assure from both looking > at the database design, and personal experience with flakey hardware, the > design is VERY failsafe and the fsck.s3ql is VERY robust about recovering > from issues. > > I found it to be effectively just as safe as straight ext4 -- I mean, the > drive that spontaneously got bad sectors, I lost one or two files that had > s3ql blocks stored on bad sectors (just as I would with ext4); if I was > writing to a s3ql when the power went out, maybe I'd lose the last couple > seconds of stuff I'd written (just as would happen with ext4.) A few > times I had things drop at the wrong moment and had to use the "sqlite3 > broken.db ".recover" | sqlite3 new.db" method to recover the database (in > local cache, it didn't back up a bad database to storage) before I could > run fsck.s3ql, but still lost nothing but the last couple seconds of data. > > In addition, it regularly backs up the database so even if the ".recover" > process had failed I would have had a good copy of the database from > earlier in the day rather than a total loss. > > Have a good day! > --Henry > On Tuesday, November 28, 2023 at 9:07:31 AM UTC-6 Paul Harris wrote: > >> On Tue, 28 Nov 2023 at 23:05, 'Georg Pfuetzenreuter' via s3ql < >> [email protected]> wrote: >> >>> >>> >>> On 11/28/23 15:24, Paul Harris wrote: >>> > Hi all, >>> > >>> > I'm confused why more people don't use backblaze's b2 ? >>> > It seems cheaper than AWS"s S3 ... >>> >>> I can also recommend checking Wasabi's S3 offering. >>> >>> >> I looked at that, it looks really good too. >> >> I just didn't like the uncertain egress/download caps. >> Would prefer to pay a known potential amount rather than annoy them and >> have them disable my account, however unlikely that could be. >> >> -- > You received this message because you are subscribed to a topic in the > Google Groups "s3ql" group. > To unsubscribe from this topic, visit > https://groups.google.com/d/topic/s3ql/403-rpd9TV4/unsubscribe. > To unsubscribe from this group and all its topics, send an email to > [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/s3ql/30044eb5-1952-4800-9f66-e8ab60a96193n%40googlegroups.com > <https://groups.google.com/d/msgid/s3ql/30044eb5-1952-4800-9f66-e8ab60a96193n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "s3ql" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/s3ql/CAMSUDdYmvTZeYxt2U2rLk5SmQ8zR0tUoCbgDasje9P%3DumzATuA%40mail.gmail.com.
