It may be helpful for those who read this thread over time to understand the definition of "fail." From time to time, TM will give users a notice that it needs to create a new backup. This is likely what "fail" means in prior posts. What's generally happened here is the TM backup on hand has got some corruption it can't repair, so it is alerting the user to the need to create a new TM backup. User can approve the prompt, it will create a new one and then it's business as usual (basically set it and forget it). Drive is fine, system is fine, fresh data backup is fine (or will be again as soon as it is finished creating the new backup).
Fail does NOT mean that all data is lost. So in a scenario where someone needed to recover from that pool of backups in that scenario, they can manually go into the files and find whatever they need to recover... almost exactly like they can in CCC or SuperDuper, etc backups. "Fail" just means how TM works- particularly in offering easy access to past versions of files- is no longer possible in the usual way... but the files are still on the disc if someone wants to dig in and find them manually... exactly as they would do in any clone-based backup software.
The big loss in this infrequent TM re-creation is the "back in time" history. Since TM is creating a brand new backup, there will be no history... exactly like simple clone backups have no history. This is akin to having an actual Time Machine able to travel back- say- 3 months MAX, now able to travel back to only earlier today... until tomorrow when it can travel back to 2 days and then the day after tomorrow when it can travel back 3 days, etc... UNTIL eventually it has the full 3 months of "back in time" capability again.
What causes this "fail"? Who knows? macOS bugs? Mac going to sleep doing a TM backup write? Gremlins? Evil Samsung trolls? Cheap Chinese chargers? <other>? But the main point is to not read more into the word than what it really means in this context.
Whatever one chooses for backup options, the smart recommendation is to always have at least TWO backups, with one stored offsite and regularly rotating with the one onsite (so both are fresh). In some scenario where fail means catastrophic loss such that one could not access files in the TM bundle at all, they could fetch the offsite TM drive and recover/restore from that one. The odds in both drives being corrupted at the same time is towards nill... even more so catastrophically corrupted. If someone is worried about being odd man out in that scenario, mix in a third drive for one more fresh backup. Each additional backup tremendously cuts the odds of data loss... especially when at least ONE is stored offsite. Guess what happens if you attach TWO TM drives and have TM use them? TM automatically alternates backups to each drive every other hour. That's also "set it and forget it" but now you have 2 independent backups (on independent hardware) going at the same time.
I've been using TM from the beginning and had no problems. Occasionally, I do get the "need to rebuild" message and by "occasionally," I mean maybe once every year or two. So I click "OK", let it rebuild (an invisible, background task while I use the Mac) and carry on normally. My offsite backup still is a full backup with history and by the time it too might need to rebuild, the onsite one has piled up "back in time" history again.
I like and have used SuperDuper and CCC at various points in the past. They are fine options as well. However, TM does seem to work quite well for me... so it would probably work well for others too... especially if they adopt the "at least 2 backup drives" (regularly rotated offsite-onsite) approach that protects against fire-flood-theft scenarios. Obviously, such scenarios that take out a Mac will also take out the lone backup drive sitting in close proximity to it (whether that one uses TM, CCC, SD or anything else).
If data is beyond precious and someone is a worst case scenario pessimist about their own data security, line up multiple drives for backup and use a combination of TM + SD + CCC with multiple drives... some stored offsite. Each additional (fresh) backup meaningfully reduces risk of data loss.
