WU has been getting slower.
Have a look in the restx.py and you will find numerous places where there are 10 second timeouts governing various connections to different services.
I have had to increase WU-related timeouts from 10 seconds to 20 seconds.
In addition I use wunderfixer regularly to catch any misses.
I did just today discover an interesting case of several false positive-missing records.
Data wasn't really missing, but was intentionally ignored by WU, because it was in between a 5-minute boundary.
In the case that I drilled into deeply a missing record was from 9:57:00 a.m., but WU already had records for 9:50, 9:54.59, and 10:00 a.m.
The wunderfixer utility defaults to 120 second time offsets, so called "epsilon" when deciding whether a record is missing. My local record from 9:57.00 is exactly 121 seconds after WU's 9:54.59 record, e.g. 1 second more than the default 120 second "epsilon," so wunderfixer deemed the record missing. I increased the epsilon to 125 seconds to compensate.
I had several such records close to the 5-minute boundaries, but just barely outside, like that. But using 125 seconds to align the buckets seems to be an effective workaround for WU showing the fractional minutes like 9:54.59, instead of just calling it 9:55.
HTH