Event Feed » History » Version 1
laforge, 06/22/2018 05:03 PM
initial page importing my text notes on ideas about an event feed
| 1 | 1 | laforge | {{>toc}} |
|---|---|---|---|
| 2 | h1. Event Feed |
||
| 3 | |||
| 4 | General idea: Feed of events pushed out from [each] Osmocom program |
||
| 5 | |||
| 6 | h2. Events |
||
| 7 | |||
| 8 | h3. FSM creation |
||
| 9 | |||
| 10 | * timestamp |
||
| 11 | * source file (string) |
||
| 12 | * source line (u32) |
||
| 13 | * ptr (u32/u64) |
||
| 14 | * fsm class (string) |
||
| 15 | * ID (string) |
||
| 16 | * parent ptr (if any) |
||
| 17 | |||
| 18 | h3. FSM event |
||
| 19 | |||
| 20 | * timestamp |
||
| 21 | * source file (string) |
||
| 22 | * source line (u32) |
||
| 23 | * ptr (u32/64) |
||
| 24 | * event number (u8/32) |
||
| 25 | * state number (u8/32) |
||
| 26 | * decoded/interpreted event payload? tlv? |
||
| 27 | |||
| 28 | h3. FSM state change |
||
| 29 | |||
| 30 | * timestamp |
||
| 31 | * source file (string) |
||
| 32 | * source line (u32) |
||
| 33 | * ptr (u32/64) |
||
| 34 | * old state number (u8/32) |
||
| 35 | * new state number (u8/32) |
||
| 36 | * timer number (int) |
||
| 37 | * timer timeout (u32) |
||
| 38 | ** optional. old timer could also remain running! |
||
| 39 | |||
| 40 | h3. FSM termination |
||
| 41 | |||
| 42 | * timestamp |
||
| 43 | * source file (string) |
||
| 44 | * source line (u32) |
||
| 45 | * ptr (u32/u64) |
||
| 46 | * termination cause |
||
| 47 | |||
| 48 | h3. FSM rename |
||
| 49 | |||
| 50 | * timestamp |
||
| 51 | * source file (string) |
||
| 52 | * source line (u32) |
||
| 53 | * ptr (u32/u64) |
||
| 54 | * old id (string) ? |
||
| 55 | * new id (string) |
||
| 56 | |||
| 57 | h3. FSM reparent (change or unlink) |
||
| 58 | |||
| 59 | * timestamp |
||
| 60 | * ptr (u32/u64) |
||
| 61 | * old parent ptr (u32/u64) |
||
| 62 | * new parent ptr (u32/u64) |
||
| 63 | |||
| 64 | h3. non-FSM events |
||
| 65 | |||
| 66 | * measurement reports (bts/bsc) |
||
| 67 | * "jolly fsm" like LAPDm, CC FSMs -> maybe convert them? |
||
| 68 | * OsmoPCU: TBF / resoure block allocations per frame number |
||
| 69 | ** probably needs some quite different handling? |
||
| 70 | |||
| 71 | |||
| 72 | h2. filtering |
||
| 73 | |||
| 74 | * ability to enable/disable generation of events per FSM class [or instance?] |
||
| 75 | * ability to do retroactive filtering in the visualization tool |
||
| 76 | |||
| 77 | h2. format / encoding |
||
| 78 | |||
| 79 | * don't generate JSON directly, let's avoid all the string encoding |
||
| 80 | * use CTRL or other interface to query string tables of FSM events and FSM states at runtime |
||
| 81 | * external program can use string tables + binary protocol to generate JSON feed for browser |
||
| 82 | * external program can collect/aggregate feeds from multiple osmocom programs |
||
| 83 | |||
| 84 | h2. visual representation |
||
| 85 | |||
| 86 | * list of instances per FSM class |
||
| 87 | ** sort it by "most recent", "ID" or "will expire soonest" |
||
| 88 | * history / time line for each FSM instance |
||
| 89 | * relationship of parent/child FSMs, maybe some kind of tree structure? |
||
| 90 | * show FSMs of different osmocom network elements in one view, i.e. BTS+BSC+MSC |
||
| 91 | |||
| 92 | |||
| 93 | h2. misc ideas |
||
| 94 | |||
| 95 | * pass out ptr/addr as u32/u64 to uniquely identify FSM |
||
| 96 | ** string ID would then only be needed during creation and rename |
||
| 97 | ** FSM class would then only be needed during creation |
||
| 98 | ** u32/u64 would be unique identifier for all FSM instances inside one osmo-* program |
||
| 99 | * generate "FSM create" events with current state at start-up (optionally?) |
