crude idea for log storage/retrieval

[mzimmers]

Hi all -

I need to start storing log messages. It would be nice if I could store ~1000 messages of up to ~100 characters each. I've already done all the work involved in routing ESP_LOG* though my own function, so now I just need a means of storing them. I was thinking of using a namespace in NVS, and store them as key value pairs. The keys would be like LOG_NNN, and I'd maintain a variable that denoted the last used value. Once I reached the limit (of 999) I'd start over, to emulate a crude ring buffer.

Before I dive into this, I just wanted to get a sense from people as to whether this idea was good/so-so/terrible. I don't think I will need to be terribly efficient (most of the time, there will just be one message every few minutes of operation, though there will be occasional bursts when a handful of messages might all come at (nearly) the same time.

Opinions? Thanks...

[tommeyers]

Good idea; logging is necessary for support-ability.

Depending upon what you see in the bursts of log entries it might help keep a more complete history if duplicates of the last entry were simply added to a counter of the last entry. It might prevent the wrap around of the log for bursts of duplicates. Depends what you see.

Tom Meyers

[gonzalo4444]

have a solution?

[mzimmers]

I did implement a logging facility, but it was more elaborate than I first thought it would be (source file was almost 1200 lines).

[ESP_igrr]

If a cloud-attached solution is an option there is now also ESP Insights. It lets you collect logs, metrics and core dumps from devices and analyze them in the cloud.

[tommeyers]

Mzimmers you work on cool stuff. I have been following your posts for years, tom