corrupted files in archives

[MonsieurNicolas]

Looking at Stellar Beat it looks like there history archives are getting corrupted somehow.

At this time, I am seeing the following corruptions (on tier 1 validators):

• truncated file error, something like Archive verification error at https://stellar-archive-3-lobstr.s3.amazonaws.com/transactions/01/6f/17/transactions-016f17ff.xdr.gz: unexpected end of file. Start repair at ledger 23999998 with [Stellar Archivist](https://github.com/stellar/go/tree/master/tools/stellar-archivist) and purge your cache afterwards.
  • LOBSTR-3 -- timestamp 2019-May-31
  • LOBSTR-4 -- timestamp 2019-Jun-1
  • LOBSTR-4 -- timestamp 2019-May-29
  • FT-SVC3 -- timestamp 2021-Apr-30
• "wrong bucket hash" error (file decompresses fine, but is empty when decompressed) Archive verification error at https://stellar-history-usw.franklintempleton.com/azuswshf401/bucket/07/57/cb/bucket-0757cbd6f482f56e176569595120d00e47c3990b63d92653a9e79233a7c2bacc.xdr.gz: Wrong bucket hash. Start repair at ledger 19999998 with [Stellar Archivist](https://github.com/stellar/go/tree/master/tools/stellar-archivist) and purge your cache afterwards.
  • FT-SCV1 -- timestamp of the faulty file is 2021-Apr-7
  • FT-SVC2 -- timestamp of the faulty file is 2021-Apr-29

Those may be two different issues, but it could be the same one (like a XYZWork problem somewhere). Truncated input files to gzip is quite confusing.

[marta-lokhova]

Note: #2211 released in Aug 2019 fixed several bugs with publishing sequence, which potentially caused some of the corrupt files described in this issue