2.1.0: Retrieve org logs, search log events, and faster analysis tools

[lcottercertinia]

What's Changed

Added

New tools

• apexlog_list_org_logs, apexlog_get_org_logs: list the debug logs in an org, filtered, sorted and paged in the org, and download them by id or the newest N (#209)
• apexlog_search_events: search a log's events in log order - what the code printed, the validation rules, or everything under one method - filtered by type, namespace or text, and paged (#144)
• apexlog_create_trace_flag: start logging a user, or set a class's or trigger's levels, for a while (#211)

New flags

• --deny-orgs: use the argument to refuse named orgs, or an org type with type:production, to every org tool. Applies to all tools that communicate with orgs. Nothing bypasses a deny, and a named org is refused before the server connects to it (#186, #209)
• --deny-org-writes: use the argument to refuse the tools that change an org - running Apex and creating trace flags - on named orgs or org types, while the tools that read its logs keep working. It takes the same entries as --deny-orgs, and nothing lifts it (#281)
• --read-only: use the argument to refuse every tool that changes an org - running Apex and creating trace flags - while the tools that read logs keep working, and list the refused tools without their parameters, ~447 tokens less on every request (#265, #276)

Existing tools

• apexlog_get_summary: report the exceptions a transaction threw, with where and how often, and how many flow elements failed (#208)
• apexlog_execute_anonymous: run from a file with apexFilePath, reuse saved scripts and save tokens (#212)

Changed

Performance

• apexlog_get_summary, apexlog_list_slow_operations, apexlog_list_limit_risks, apexlog_search_events: answer the first call on a log of 1 MB or more 2.5 to 8 times faster, and hold 3.5 to 5.5 times less memory for that log, on a log parser that reads it 4 to 8 times faster (#246, #277)
  • apexlog_get_summary: answer repeat calls on the same log 8 to 25 times faster
  • apexlog_search_events: answer repeat calls on the same log 2.5 to 11 times faster
  • apexlog_list_slow_operations: answer repeat calls on the same log 1.5 to 2.5 times faster

Log analysis

• apexlog_get_summary, apexlog_list_slow_operations, apexlog_list_limit_risks, apexlog_search_events: read only the first log of a file that holds several, and say in warning how many it holds, instead of reading them all as one (#278)
• apexlog_get_summary: make the categories add up to the whole log. The transaction's own time counts under its category, and a new unattributed row holds the time no event spans (#226)
• apexlog_get_summary, apexlog_list_slow_operations, apexlog_execute_anonymous: count the log's duration from its first line, not from the start of the transaction (#226)
• apexlog_list_slow_operations: give nextOffset, the offset of the next page, instead of leaving you to count the rows you got (#248)

Running Apex

• apexlog_execute_anonymous: run at your trace flag's levels when you give no debugLevel, else at the defaults, and set the categories an object debugLevel names over the defaults - never over the levels a previous run left (#230)
• apexlog_execute_anonymous: rename durationMs to durationTotalMs (#240)
• apexlog_execute_anonymous, apexlog_create_trace_flag: end a long reason in a production refusal or confirmation with "…", and never split a character (#272)

Flags

• --no-apex-execution: list apexlog_execute_anonymous without its parameters, which no call can use, ~257 tokens less on every request (#276)

Fixed

Every tool

• every tool: refuse a parameter it does not have, instead of ignoring it and running without it (#244)

Running Apex

• apexlog_execute_anonymous: stop changing your trace flag's debug level or leaving a 24-hour trace flag on your user (#207)
• apexlog_execute_anonymous: run the Apex with a warning when the org refuses a trace flag or its debug level, instead of failing the call (#207, #269)
• apexlog_execute_anonymous: refuse an outputDir it cannot write before running the Apex, instead of failing after its DML has committed (#260)
• apexlog_execute_anonymous: show all of the Apex in a production confirmation, and refuse Apex too long to show (#221)
• apexlog_execute_anonymous: show the log levels in a production confirmation - a live trace flag's where it outranks debugLevel - and refuse a run at levels other than the ones confirmed (#228, #230)
• apexlog_execute_anonymous: stop a 60-second stall waiting on roots (#222)
• apexlog_execute_anonymous: require targetOrg when the client's roots can't be used (#225)

Log analysis

• apexlog_get_summary, apexlog_list_slow_operations: time a method that the log never closes up to the end of the log, instead of reporting it too short (#226)
• apexlog_list_slow_operations: give the cost of a query plan that read a skinny table, instead of null (#279)

Error messages

• apexlog_execute_anonymous: say when the org's login has expired and how to log in again (#239)
• apexlog_execute_anonymous: say when the org's log storage is full and where to free it, instead of only that the trace flag was refused (#245)
• apexlog_execute_anonymous: say when targetOrg names no authenticated org, instead of the Salesforce SDK's raw error (#236)
• apexlog_get_summary, apexlog_list_slow_operations, apexlog_list_limit_risks, apexlog_search_events: name the log and the cause when it cannot be read, such as a log larger than 2 GiB, instead of Node's raw error (#277)
• apexlog_get_summary, apexlog_list_slow_operations, apexlog_list_limit_risks: say a logFilePath that is a directory is one, instead of Node's raw error (#234)

Full Changelog: 2.0.1...2.1.0

---

This discussion was created from the release 2.1.0: Retrieve org logs, search log events, and faster analysis tools.