| 11:57 |
whaack |
I cut the power to my bitcoind abruptly and on startup saw this NSt8ios_base7failureE in my debug file. http://paste.deedbot.org/?id=5ls_ From searching logs it seems I gotta start from scratch or do some dumpblock /eatblock work |
| 11:57 |
whaack |
cut the power to my machine running bitcoind* |
| |
~ 38 minutes ~ |
| 12:36 |
asciilifeform |
whaack: seems that your db is corrupted. |
| |
~ 23 minutes ~ |
| 13:00 |
whaack |
asciilifeform: yup, just used your cutblock tool, very nifty, figuring out how to clear my db and feed it back in atm |
| 13:11 |
asciilifeform |
whaack: interestingly, in all of asciilifeform's time w/ trb, never personally saw the db die. (plenty of folx reported, tho, very similarly to yours) |
| 13:12 |
asciilifeform |
whaack: outta curiosity, was this a box w/ traditional mechanical hdd ? |
| 13:13 |
whaack |
asciilifeform: no, I am writing to an ssd |
| 13:16 |
asciilifeform |
interesting. i had assumed that never witnessed hosed db on acct of using ssd. |
| 13:17 |
* |
asciilifeform wonders if also depends on the particular point in a node sync |
| 13:17 |
asciilifeform |
whaack: was this a fully-synced noad ? |
| 13:17 |
whaack |
asciilifeform: nope, it was at block height 213777 |
| 13:20 |
* |
asciilifeform never had chance to investigate how db corrupts, because never 1ce reproduced. despite plenty (most recently -- yesterday..) times when noad lost power abruptly; or process kill -9'd; and similar |
| 13:30 |
whaack |
Here were the last few lines from debug.log before I saw the corrupted db error http://paste.deedbot.org/?id=keIr . Looks like the program was writing to debug.log and only got to the "E" in Error when the power cut |
| 13:33 |
whaack |
db.log was empty, not sure if that is the norm |
| 13:35 |
asciilifeform |
whaack: outta curiosity, what filesystem in use on that machine ? |
| 13:43 |
whaack |
ext4 |
| 13:43 |
asciilifeform |
so, same as in use on my nodes. |