out_forward: fix memory leak during connection loss

[Garfield96]

I observed that memory consumption was constantly growing while the data sink of the forward plugin was unavailable. Using valgrind, it could be confirmed that there is a leak in the forward output plugin:

==8== 385,024 bytes in 47 blocks are definitely lost in loss record 126 of 126
==8==    at 0x4E055B4: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8==    by 0x4E0A81C: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==8==    by 0x739B7B: msgpack_sbuffer_write (sbuffer.h:81)
==8==    by 0x5D5262: msgpack_pack_map (pack_template.h:753)
==8==    by 0x5D8F8A: flb_mp_map_header_init (flb_mp.c:321)
==8==    by 0x73A026: append_options (forward_format.c:103)
==8==    by 0x73AE2F: flb_forward_format_forward_mode (forward_format.c:442)
==8==    by 0x73B3A4: flb_forward_format (forward_format.c:633)
==8==    by 0x73566E: cb_forward_flush (forward.c:1559)
==8==    by 0x51F033: output_pre_cb_flush (flb_output.h:597)
==8==    by 0xA45126: co_init (amd64.c:117)

Further analysis showed that during connection loss out_buf is only freed if time_as_integer is set to true:

        if (!u_conn) {
            flb_plg_error(ctx->ins, "no upstream connections available");
            msgpack_sbuffer_destroy(&mp_sbuf);
            if (fc->time_as_integer == FLB_TRUE) {
                flb_free(out_buf);
            }

Source: https://github.com/fluent/fluent-bit/blob/master/plugins/out_forward/forward.c#L1574-L1579

However, from the code it looks like out_buf is also used to write metadata in append_options (https://github.com/fluent/fluent-bit/blob/master/plugins/out_forward/forward_format.c#L85). To verify this assumption, I enabled Time_as_Integer in the forward configuration. Afterwards, valgrind did no longer find a leak. Therefore, I removed the conditional, which is save because this is also done in other places (e.g. https://github.com/fluent/fluent-bit/blob/master/plugins/out_forward/forward.c#L1644) and even if out_buf is not used, it would be NULL and freeing NULL is valid.

I ran the patched version for 12 hours and the memory consumption was perfectly constant.

---

Enter [N/A] in the box, if an item is not applicable to your change.

Testing
Before we can approve your change; please submit the following in a comment:

• [N/A] Example configuration file for the change -> Can be reproduced with any forward configuration which doesn't set Time_as_Integer to true
• [N/A] Debug log output from testing the change -> Change not visible in log

• Attached Valgrind output that shows no leaks or memory corruption was found -> Nothing to attach, since valgrind found only known issues

If this is a change to packaging of containers or native binaries then please confirm it works for all targets.

• [N/A] Run local packaging test showing all targets (including any new ones) build.
• [N/A] Set ok-package-test label to test for all targets (requires maintainer to do).

Documentation

• [N/A] Documentation required for this feature

Backporting

• Backport to latest stable release.

---

Fluent Bit is licensed under Apache 2.0, by submitting this pull request I understand that this code will be released under the terms of that license.

[edsiper]

good catch! thanks for the PR