简体中文界面。帖子、指南及政策正文保留原文；未对原文作自动翻译。
KaiBrooks0802 · 2026年1月23日 10:40 UTC
Time request start and return separately from parsing, file writing and the display update. Don't put that timing trace through the same writer you're investigating, or it can tell you everything happened late simply because it couldn't tell you anything earlier.
13分The separate trace shows the reply returning promptly, then the display waiting for the file write; we can reproduce the gap with a deliberately delayed test writer using saved responses
13分What happens to the displayed reading's age during that pause?
14分And does a failed status read have a separate test? Fixing the writer shouldn't make a missing reply look like a current value.
0分Maya and Rachel, both matter. I wouldn't solve this with an unlimited queue either. A folder can stay unavailable longer than the test, and somebody has to decide what the monitor does when it can't keep every diagnostic record.
10分Maintainer and maintenance lead agreed bounded buffering with an explicit recording-gap indication, while status age remains beside the latest valid reading; failed reads leave that reading visibly old rather than reporting a fresh success
18分Have they kept observation time separate from the later write time in the saved records? Otherwise the faster screen can leave a misleading trend file behind it.
15分Yes, saved-response replay preserves observation times while delaying writes, and full-buffer cases show the agreed gap; failed reads are covered separately, including startup without any valid reading
19分Has the normal installed monitor been checked with the change now?
12分Installed read-only comparison completed: display stays responsive during the controlled writer delay, shows reading age correctly and flags the tested recording gap; maintenance has the buffer policy and failed-read cases, so we've closed the logging stall fix
6分此讨论已连续六个月没有活动，现已关闭新回复。 最后活动： .