You might find this useful:
https://lucidworks.com/2013/08/23/understanding-transaction-logs-softcommit-and-commit-in-sorlcloud/
bq: So I can never have soft auto commit after each update ? This
sounds like a bug to me.
No, rather a fundamental misunderstanding of the difference between an
ACID system and Solr.
Best,
Erick
On Fri, Aug 25, 2017 at 10:08 AM, Shawn Heisey <apa...@elyograg.org> wrote:
> On 8/24/2017 4:59 AM, Angel Todorov wrote:
>> I also tested, of course, by setting a value of 0, expecting that it would
>> work in the way I expect it to , but unfortunately - it doesn't. Nothing is
>> committed in that case.
>
> Settings of zero turn that part of the automatic commit off. That's a
> way for people to have the setting in their config but not have the
> setting active.
>
> I think I may have found the off-by-one problem in the CommitTracker
> code. In the master branch, line 161 reads:
>
> if (docs == docsUpperBound + 1) {
>
> Based on how the source object for the "docs" variable is handled, I
> think it should probably be:
>
> if (docs >= docsUpperBound) {
>
> While what you have seen does look like a bug to me, and something that
> I think should be fixed, configuring values that low is generally a bad
> idea. Commits that open a new searcher require a fair amount of time
> and cpu/memory resources to complete -- even soft commits. They should
> not be configured to happen extremely frequently, which a single-digit
> maxDocs value will do. That is likely to cause overlapping commits,
> which can quickly become a major performance issue.
>
> Configuring only maxTime usually results in more predictable operation
> than maxDocs.
>
> Thanks,
> Shawn
>
