Why some searches take longer
Anything already in the index is instant. A search nobody has run recently has to ask the boards live, and the slowest one sets the pace.
A search we already hold results for returns almost instantly. One nobody has run recently has to be asked live, and the slowest board sets the pace for the whole thing.
- Rows stream in as they arrive rather than waiting for the last board. The list grows for a few seconds.
- The progress bar shows how many boards have answered and how many jobs are in so far.
- You can stop it at any point and keep everything that already arrived.
- The same search again is faster. A search served from the index also quietly refreshes itself in the background, so the second run is both quicker and newer.
A live search does not ask every board we cover. It asks the source that leads in that country plus the general aggregators, which is usually five to seven of them, and a remote search adds the remote-only boards on top.
Where it stops: We cannot make a slow board fast. If one source is having a bad day the search still completes without it, and the failure is recorded rather than hidden. See when a job board fails.
Related
- When a job board failsOne source failing does not fail the search. The others still answer, and the failure is recorded rather than hidden.
- How fresh the results areThere is no single refresh interval. Popular searches are refreshed far more often than obscure ones, and running one again makes it fresher.
- When a search returns nothingEither your filters removed everything, in which case the page names the one responsible, or the index holds nothing yet for that query.
Did this answer your question?
Still stuck? Ask us directly, and mention this article so we know where to start.