checkpoint_start processing is failing during load test

[hibari]

checkpoint_start processing is failing during load test

[kinogmt]

Can you briefly explain how to reproduce this issue?

[norton]

$ cd workdir-name/hibari/hibari
$ repo sync
$ make check-package

995e742

[kinogmt]

This looks like a regression at the following commit.

hibari/gdss-brick@c109249

it used to ignore enoent errors returned from brick_ets:delete_external_bad_sequence_file/1.
But it's now matching with 'ok' and crashing when the file does not exist.

ok = delete_external_bad_sequence_file(S_ro#state.name),

-----------------------------------------------------------------
[kino@bw hibari] bin/hibari start
[kino@bw hibari] bin/hibari attach
Attaching to /tmp//home/kino/work/hibari/hibari/tmp/hibari/erlang.pipe.1 (^D to exit)

Eshell V5.8.2  (abort with ^G)
([email protected])1> gmt_util:dbgon(brick_ets,delete_external_bad_sequence_file),brick_server:checkpoint_bricks_and_wait('[email protected]').
(<0.325.0>) call brick_ets:delete_external_bad_sequence_file(tab1_ch1_b1)
(<0.325.0>) returned from brick_ets:delete_external_bad_sequence_file/1 -> {error,
                                                                            enoent}

=ERROR REPORT==== 20-Jan-2011::11:47:59 ===
Error in process <0.325.0> on node '[email protected]' with exit value: {{badmatch,{error,enoent}},[{brick_ets,checkpoint_start,4}]}


=ERROR REPORT==== 20-Jan-2011::11:47:59 ===
    module: brick_ets
    line: 512
    msg: "checkpoint: pid <0.325.0> died with {{badmatch,{error,enoent}},\n                                     [{brick_ets,checkpoint_start,4}]}\n"
(<0.328.0>) call brick_ets:delete_external_bad_sequence_file(bootstrap_copy1)
(<0.328.0>) returned from brick_ets:delete_external_bad_sequence_file/1 -> {error,
                                                                            enoent}

[kinogmt]

Fixed.

hibari/gdss-brick@1250a16
hibari/gdss-brick@0d687bb